Android Camera HAL3.2 Properties

Table of Contents

Properties

Property Name Type Description Units Range Tags
colorCorrection
controls
Property Name Type Description Units Range Tags
android.colorCorrection.mode byte [public] [full]
  • TRANSFORM_MATRIX

    Use the android.colorCorrection.transform matrix and android.colorCorrection.gains to do color conversion.

    All advanced white balance adjustments (not specified by our white balance pipeline) must be disabled.

    If AWB is enabled with android.control.awbMode != OFF, then TRANSFORM_MATRIX is ignored. The camera device will override this value to either FAST or HIGH_QUALITY.

  • FAST

    Color correction processing must not slow down capture rate relative to sensor raw output.

    Advanced white balance adjustments above and beyond the specified white balance pipeline may be applied.

    If AWB is enabled with android.control.awbMode != OFF, then the camera device uses the last frame's AWB values (or defaults if AWB has never been run).

  • HIGH_QUALITY

    Color correction processing operates at improved quality but the capture rate might be reduced (relative to sensor raw output rate)

    Advanced white balance adjustments above and beyond the specified white balance pipeline may be applied.

    If AWB is enabled with android.control.awbMode != OFF, then the camera device uses the last frame's AWB values (or defaults if AWB has never been run).

The mode control selects how the image data is converted from the sensor's native color into linear sRGB color.

Details

When auto-white balance (AWB) is enabled with android.control.awbMode, this control is overridden by the AWB routine. When AWB is disabled, the application controls how the color mapping is performed.

We define the expected processing pipeline below. For consistency across devices, this is always the case with TRANSFORM_MATRIX.

When either FULL or HIGH_QUALITY is used, the camera device may do additional processing but android.colorCorrection.gains and android.colorCorrection.transform will still be provided by the camera device (in the results) and be roughly correct.

Switching to TRANSFORM_MATRIX and using the data provided from FAST or HIGH_QUALITY will yield a picture with the same white point as what was produced by the camera device in the earlier frame.

The expected processing pipeline is as follows:

White balance processing pipeline

The white balance is encoded by two values, a 4-channel white-balance gain vector (applied in the Bayer domain), and a 3x3 color transform matrix (applied after demosaic).

The 4-channel white-balance gains are defined as:

android.colorCorrection.gains = [ R G_even G_odd B ]

where G_even is the gain for green pixels on even rows of the output, and G_odd is the gain for green pixels on the odd rows. These may be identical for a given camera device implementation; if the camera device does not support a separate gain for even/odd green channels, it will use the G_even value, and write G_odd equal to G_even in the output result metadata.

The matrices for color transforms are defined as a 9-entry vector:

android.colorCorrection.transform = [ I0 I1 I2 I3 I4 I5 I6 I7 I8 ]

which define a transform from input sensor colors, P_in = [ r g b ], to output linear sRGB, P_out = [ r' g' b' ],

with colors as follows:

r' = I0r + I1g + I2b
g' = I3r + I4g + I5b
b' = I6r + I7g + I8b

Both the input and output value ranges must match. Overflow/underflow values are clipped to fit within the range.

HAL Implementation Details

HAL must support both FAST and HIGH_QUALITY if color correction control is available on the camera device, but the underlying implementation can be the same for both modes. That is, if the highest quality implementation on the camera device does not slow down capture rate, then FAST and HIGH_QUALITY should generate the same output.

android.colorCorrection.transform rational x 3 x 3 [public as colorSpaceTransform] [full]
3x3 rational matrix in row-major order

A color transform matrix to use to transform from sensor RGB color space to output linear sRGB color space.

Unitless scale factors
Details

This matrix is either set by the camera device when the request android.colorCorrection.mode is not TRANSFORM_MATRIX, or directly by the application in the request when the android.colorCorrection.mode is TRANSFORM_MATRIX.

In the latter case, the camera device may round the matrix to account for precision issues; the final rounded matrix should be reported back in this matrix result metadata. The transform should keep the magnitude of the output color values within [0, 1.0] (assuming input color values is within the normalized range [0, 1.0]), or clipping may occur.

The valid range of each matrix element varies on different devices, but values within [-1.5, 3.0] are guaranteed not to be clipped.

android.colorCorrection.gains float x 4 [public as rggbChannelVector] [full]
A 1D array of floats for 4 color channel gains

Gains applying to Bayer raw color channels for white-balance.

Unitless gain factors
Details

These per-channel gains are either set by the camera device when the request android.colorCorrection.mode is not TRANSFORM_MATRIX, or directly by the application in the request when the android.colorCorrection.mode is TRANSFORM_MATRIX.

The gains in the result metadata are the gains actually applied by the camera device to the current frame.

The valid range of gains varies on different devices, but gains between [1.0, 3.0] are guaranteed not to be clipped. Even if a given device allows gains below 1.0, this is usually not recommended because this can create color artifacts.

HAL Implementation Details

The 4-channel white-balance gains are defined in the order of [R G_even G_odd B], where G_even is the gain for green pixels on even rows of the output, and G_odd is the gain for green pixels on the odd rows.

If a HAL does not support a separate gain for even/odd green channels, it must use the G_even value, and write G_odd equal to G_even in the output result metadata.

android.colorCorrection.aberrationMode byte [public] [legacy]
  • OFF

    No aberration correction is applied.

  • FAST

    Aberration correction will not slow down capture rate relative to sensor raw output.

  • HIGH_QUALITY

    Aberration correction operates at improved quality but the capture rate might be reduced (relative to sensor raw output rate)

Mode of operation for the chromatic aberration correction algorithm.

android.colorCorrection.availableAberrationModes

Details

Chromatic (color) aberration is caused by the fact that different wavelengths of light can not focus on the same point after exiting from the lens. This metadata defines the high level control of chromatic aberration correction algorithm, which aims to minimize the chromatic artifacts that may occur along the object boundaries in an image.

FAST/HIGH_QUALITY both mean that camera device determined aberration correction will be applied. HIGH_QUALITY mode indicates that the camera device will use the highest-quality aberration correction algorithms, even if it slows down capture rate. FAST means the camera device will not slow down capture rate when applying aberration correction.

LEGACY devices will always be in FAST mode.

dynamic
Property Name Type Description Units Range Tags
android.colorCorrection.mode byte [public] [full]
  • TRANSFORM_MATRIX

    Use the android.colorCorrection.transform matrix and android.colorCorrection.gains to do color conversion.

    All advanced white balance adjustments (not specified by our white balance pipeline) must be disabled.

    If AWB is enabled with android.control.awbMode != OFF, then TRANSFORM_MATRIX is ignored. The camera device will override this value to either FAST or HIGH_QUALITY.

  • FAST

    Color correction processing must not slow down capture rate relative to sensor raw output.

    Advanced white balance adjustments above and beyond the specified white balance pipeline may be applied.

    If AWB is enabled with android.control.awbMode != OFF, then the camera device uses the last frame's AWB values (or defaults if AWB has never been run).

  • HIGH_QUALITY

    Color correction processing operates at improved quality but the capture rate might be reduced (relative to sensor raw output rate)

    Advanced white balance adjustments above and beyond the specified white balance pipeline may be applied.

    If AWB is enabled with android.control.awbMode != OFF, then the camera device uses the last frame's AWB values (or defaults if AWB has never been run).

The mode control selects how the image data is converted from the sensor's native color into linear sRGB color.

Details

When auto-white balance (AWB) is enabled with android.control.awbMode, this control is overridden by the AWB routine. When AWB is disabled, the application controls how the color mapping is performed.

We define the expected processing pipeline below. For consistency across devices, this is always the case with TRANSFORM_MATRIX.

When either FULL or HIGH_QUALITY is used, the camera device may do additional processing but android.colorCorrection.gains and android.colorCorrection.transform will still be provided by the camera device (in the results) and be roughly correct.

Switching to TRANSFORM_MATRIX and using the data provided from FAST or HIGH_QUALITY will yield a picture with the same white point as what was produced by the camera device in the earlier frame.

The expected processing pipeline is as follows:

White balance processing pipeline

The white balance is encoded by two values, a 4-channel white-balance gain vector (applied in the Bayer domain), and a 3x3 color transform matrix (applied after demosaic).

The 4-channel white-balance gains are defined as:

android.colorCorrection.gains = [ R G_even G_odd B ]

where G_even is the gain for green pixels on even rows of the output, and G_odd is the gain for green pixels on the odd rows. These may be identical for a given camera device implementation; if the camera device does not support a separate gain for even/odd green channels, it will use the G_even value, and write G_odd equal to G_even in the output result metadata.

The matrices for color transforms are defined as a 9-entry vector:

android.colorCorrection.transform = [ I0 I1 I2 I3 I4 I5 I6 I7 I8 ]

which define a transform from input sensor colors, P_in = [ r g b ], to output linear sRGB, P_out = [ r' g' b' ],

with colors as follows:

r' = I0r + I1g + I2b
g' = I3r + I4g + I5b
b' = I6r + I7g + I8b

Both the input and output value ranges must match. Overflow/underflow values are clipped to fit within the range.

HAL Implementation Details

HAL must support both FAST and HIGH_QUALITY if color correction control is available on the camera device, but the underlying implementation can be the same for both modes. That is, if the highest quality implementation on the camera device does not slow down capture rate, then FAST and HIGH_QUALITY should generate the same output.

android.colorCorrection.transform rational x 3 x 3 [public as colorSpaceTransform] [full]
3x3 rational matrix in row-major order

A color transform matrix to use to transform from sensor RGB color space to output linear sRGB color space.

Unitless scale factors
Details

This matrix is either set by the camera device when the request android.colorCorrection.mode is not TRANSFORM_MATRIX, or directly by the application in the request when the android.colorCorrection.mode is TRANSFORM_MATRIX.

In the latter case, the camera device may round the matrix to account for precision issues; the final rounded matrix should be reported back in this matrix result metadata. The transform should keep the magnitude of the output color values within [0, 1.0] (assuming input color values is within the normalized range [0, 1.0]), or clipping may occur.

The valid range of each matrix element varies on different devices, but values within [-1.5, 3.0] are guaranteed not to be clipped.

android.colorCorrection.gains float x 4 [public as rggbChannelVector] [full]
A 1D array of floats for 4 color channel gains

Gains applying to Bayer raw color channels for white-balance.

Unitless gain factors
Details

These per-channel gains are either set by the camera device when the request android.colorCorrection.mode is not TRANSFORM_MATRIX, or directly by the application in the request when the android.colorCorrection.mode is TRANSFORM_MATRIX.

The gains in the result metadata are the gains actually applied by the camera device to the current frame.

The valid range of gains varies on different devices, but gains between [1.0, 3.0] are guaranteed not to be clipped. Even if a given device allows gains below 1.0, this is usually not recommended because this can create color artifacts.

HAL Implementation Details

The 4-channel white-balance gains are defined in the order of [R G_even G_odd B], where G_even is the gain for green pixels on even rows of the output, and G_odd is the gain for green pixels on the odd rows.

If a HAL does not support a separate gain for even/odd green channels, it must use the G_even value, and write G_odd equal to G_even in the output result metadata.

android.colorCorrection.aberrationMode byte [public] [legacy]
  • OFF

    No aberration correction is applied.

  • FAST

    Aberration correction will not slow down capture rate relative to sensor raw output.

  • HIGH_QUALITY

    Aberration correction operates at improved quality but the capture rate might be reduced (relative to sensor raw output rate)

Mode of operation for the chromatic aberration correction algorithm.

android.colorCorrection.availableAberrationModes

Details

Chromatic (color) aberration is caused by the fact that different wavelengths of light can not focus on the same point after exiting from the lens. This metadata defines the high level control of chromatic aberration correction algorithm, which aims to minimize the chromatic artifacts that may occur along the object boundaries in an image.

FAST/HIGH_QUALITY both mean that camera device determined aberration correction will be applied. HIGH_QUALITY mode indicates that the camera device will use the highest-quality aberration correction algorithms, even if it slows down capture rate. FAST means the camera device will not slow down capture rate when applying aberration correction.

LEGACY devices will always be in FAST mode.

static
Property Name Type Description Units Range Tags
android.colorCorrection.availableAberrationModes byte x n [public as enumList] [legacy]
list of enums

List of aberration correction modes for android.colorCorrection.aberrationMode that are supported by this camera device.

Any value listed in android.colorCorrection.aberrationMode

Details

This key lists the valid modes for android.colorCorrection.aberrationMode. If no aberration correction modes are available for a device, this list will solely include OFF mode. All camera devices will support either OFF or FAST mode.

Camera devices that support the MANUAL_POST_PROCESSING capability will always list OFF mode. This includes all FULL level devices.

LEGACY devices will always only support FAST mode.

HAL Implementation Details

HAL must support both FAST and HIGH_QUALITY if chromatic aberration control is available on the camera device, but the underlying implementation can be the same for both modes. That is, if the highest quality implementation on the camera device does not slow down capture rate, then FAST and HIGH_QUALITY will generate the same output.

control
controls
Property Name Type Description Units Range Tags
android.control.aeAntibandingMode byte [public] [legacy]
  • OFF

    The camera device will not adjust exposure duration to avoid banding problems.

  • 50HZ

    The camera device will adjust exposure duration to avoid banding problems with 50Hz illumination sources.

  • 60HZ

    The camera device will adjust exposure duration to avoid banding problems with 60Hz illumination sources.

  • AUTO

    The camera device will automatically adapt its antibanding routine to the current illumination condition. This is the default mode if AUTO is available on given camera device.

The desired setting for the camera device's auto-exposure algorithm's antibanding compensation.

android.control.aeAvailableAntibandingModes

Details

Some kinds of lighting fixtures, such as some fluorescent lights, flicker at the rate of the power supply frequency (60Hz or 50Hz, depending on country). While this is typically not noticeable to a person, it can be visible to a camera device. If a camera sets its exposure time to the wrong value, the flicker may become visible in the viewfinder as flicker or in a final captured image, as a set of variable-brightness bands across the image.

Therefore, the auto-exposure routines of camera devices include antibanding routines that ensure that the chosen exposure value will not cause such banding. The choice of exposure time depends on the rate of flicker, which the camera device can detect automatically, or the expected rate can be selected by the application using this control.

A given camera device may not support all of the possible options for the antibanding mode. The android.control.aeAvailableAntibandingModes key contains the available modes for a given camera device.

AUTO mode is the default if it is available on given camera device. When AUTO mode is not available, the default will be either 50HZ or 60HZ, and both 50HZ and 60HZ will be available.

If manual exposure control is enabled (by setting android.control.aeMode or android.control.mode to OFF), then this setting has no effect, and the application must ensure it selects exposure times that do not cause banding issues. The android.statistics.sceneFlicker key can assist the application in this.

HAL Implementation Details

For all capture request templates, this field must be set to AUTO if AUTO mode is available. If AUTO is not available, the default must be either 50HZ or 60HZ, and both 50HZ and 60HZ must be available.

If manual exposure control is enabled (by setting android.control.aeMode or android.control.mode to OFF), then the exposure values provided by the application must not be adjusted for antibanding.

android.control.aeExposureCompensation int32 [public] [legacy]

Adjustment to auto-exposure (AE) target image brightness.

Compensation steps

android.control.aeCompensationRange

Details

The adjustment is measured as a count of steps, with the step size defined by android.control.aeCompensationStep and the allowed range by android.control.aeCompensationRange.

For example, if the exposure value (EV) step is 0.333, '6' will mean an exposure compensation of +2 EV; -3 will mean an exposure compensation of -1 EV. One EV represents a doubling of image brightness. Note that this control will only be effective if android.control.aeMode != OFF. This control will take effect even when android.control.aeLock == true.

In the event of exposure compensation value being changed, camera device may take several frames to reach the newly requested exposure target. During that time, android.control.aeState field will be in the SEARCHING state. Once the new exposure target is reached, android.control.aeState will change from SEARCHING to either CONVERGED, LOCKED (if AE lock is enabled), or FLASH_REQUIRED (if the scene is too dark for still capture).

android.control.aeLock byte [public as boolean] [legacy]
  • OFF

    Auto-exposure lock is disabled; the AE algorithm is free to update its parameters.

  • ON

    Auto-exposure lock is enabled; the AE algorithm must not update the exposure and sensitivity parameters while the lock is active.

    android.control.aeExposureCompensation setting changes will still take effect while auto-exposure is locked.

    Some rare LEGACY devices may not support this, in which case the value will always be overridden to OFF.

Whether auto-exposure (AE) is currently locked to its latest calculated values.

Details

When set to true (ON), the AE algorithm is locked to its latest parameters, and will not change exposure settings until the lock is set to false (OFF).

Note that even when AE is locked, the flash may be fired if the android.control.aeMode is ON_AUTO_FLASH / ON_ALWAYS_FLASH / ON_AUTO_FLASH_REDEYE.

When android.control.aeExposureCompensation is changed, even if the AE lock is ON, the camera device will still adjust its exposure value.

If AE precapture is triggered (see android.control.aePrecaptureTrigger) when AE is already locked, the camera device will not change the exposure time (android.sensor.exposureTime) and sensitivity (android.sensor.sensitivity) parameters. The flash may be fired if the android.control.aeMode is ON_AUTO_FLASH/ON_AUTO_FLASH_REDEYE and the scene is too dark. If the android.control.aeMode is ON_ALWAYS_FLASH, the scene may become overexposed. Similarly, AE precapture trigger CANCEL has no effect when AE is already locked.

When an AE precapture sequence is triggered, AE unlock will not be able to unlock the AE if AE is locked by the camera device internally during precapture metering sequence In other words, submitting requests with AE unlock has no effect for an ongoing precapture metering sequence. Otherwise, the precapture metering sequence will never succeed in a sequence of preview requests where AE lock is always set to false.

Since the camera device has a pipeline of in-flight requests, the settings that get locked do not necessarily correspond to the settings that were present in the latest capture result received from the camera device, since additional captures and AE updates may have occurred even before the result was sent out. If an application is switching between automatic and manual control and wishes to eliminate any flicker during the switch, the following procedure is recommended:

  1. Starting in auto-AE mode:
  2. Lock AE
  3. Wait for the first result to be output that has the AE locked
  4. Copy exposure settings from that result into a request, set the request to manual AE
  5. Submit the capture request, proceed to run manual AE as desired.

See android.control.aeState for AE lock related state transition details.

android.control.aeMode byte [public] [legacy]
  • OFF

    The camera device's autoexposure routine is disabled.

    The application-selected android.sensor.exposureTime, android.sensor.sensitivity and android.sensor.frameDuration are used by the camera device, along with android.flash.* fields, if there's a flash unit for this camera device.

    Note that auto-white balance (AWB) and auto-focus (AF) behavior is device dependent when AE is in OFF mode. To have consistent behavior across different devices, it is recommended to either set AWB and AF to OFF mode or lock AWB and AF before setting AE to OFF. See android.control.awbMode, android.control.afMode, android.control.awbLock, and android.control.afTrigger for more details.

    LEGACY devices do not support the OFF mode and will override attempts to use this value to ON.

  • ON

    The camera device's autoexposure routine is active, with no flash control.

    The application's values for android.sensor.exposureTime, android.sensor.sensitivity, and android.sensor.frameDuration are ignored. The application has control over the various android.flash.* fields.

  • ON_AUTO_FLASH

    Like ON, except that the camera device also controls the camera's flash unit, firing it in low-light conditions.

    The flash may be fired during a precapture sequence (triggered by android.control.aePrecaptureTrigger) and may be fired for captures for which the android.control.captureIntent field is set to STILL_CAPTURE

  • ON_ALWAYS_FLASH

    Like ON, except that the camera device also controls the camera's flash unit, always firing it for still captures.

    The flash may be fired during a precapture sequence (triggered by android.control.aePrecaptureTrigger) and will always be fired for captures for which the android.control.captureIntent field is set to STILL_CAPTURE

  • ON_AUTO_FLASH_REDEYE

    Like ON_AUTO_FLASH, but with automatic red eye reduction.

    If deemed necessary by the camera device, a red eye reduction flash will fire during the precapture sequence.

The desired mode for the camera device's auto-exposure routine.

android.control.aeAvailableModes

Details

This control is only effective if android.control.mode is AUTO.

When set to any of the ON modes, the camera device's auto-exposure routine is enabled, overriding the application's selected exposure time, sensor sensitivity, and frame duration (android.sensor.exposureTime, android.sensor.sensitivity, and android.sensor.frameDuration). If one of the FLASH modes is selected, the camera device's flash unit controls are also overridden.

The FLASH modes are only available if the camera device has a flash unit (android.flash.info.available is true).

If flash TORCH mode is desired, this field must be set to ON or OFF, and android.flash.mode set to TORCH.

When set to any of the ON modes, the values chosen by the camera device auto-exposure routine for the overridden fields for a given capture will be available in its CaptureResult.

android.control.aeRegions int32 x 5 x area_count [public as meteringRectangle]

List of metering areas to use for auto-exposure adjustment.

Pixel coordinates within android.sensor.info.activeArraySize

Coordinates must be between [(0,0), (width, height)) of android.sensor.info.activeArraySize

Details

Not available if android.control.maxRegionsAe is 0. Otherwise will always be present.

The maximum number of regions supported by the device is determined by the value of android.control.maxRegionsAe.

The coordinate system is based on the active pixel array, with (0,0) being the top-left pixel in the active pixel array, and (android.sensor.info.activeArraySize.width - 1, android.sensor.info.activeArraySize.height - 1) being the bottom-right pixel in the active pixel array.

The weight must be within [0, 1000], and represents a weight for every pixel in the area. This means that a large metering area with the same weight as a smaller area will have more effect in the metering result. Metering areas can partially overlap and the camera device will add the weights in the overlap region.

The weights are relative to weights of other exposure metering regions, so if only one region is used, all non-zero weights will have the same effect. A region with 0 weight is ignored.

If all regions have 0 weight, then no specific metering area needs to be used by the camera device.

If the metering region is outside the used android.scaler.cropRegion returned in capture result metadata, the camera device will ignore the sections outside the crop region and output only the intersection rectangle as the metering region in the result metadata. If the region is entirely outside the crop region, it will be ignored and not reported in the result metadata.

HAL Implementation Details

The HAL level representation of MeteringRectangle[] is a int[5 * area_count]. Every five elements represent a metering region of (xmin, ymin, xmax, ymax, weight). The rectangle is defined to be inclusive on xmin and ymin, but exclusive on xmax and ymax.

android.control.aeTargetFpsRange int32 x 2 [public as rangeInt] [legacy]

Range over which the auto-exposure routine can adjust the capture frame rate to maintain good exposure.

Frames per second (FPS)

Any of the entries in android.control.aeAvailableTargetFpsRanges

Details

Only constrains auto-exposure (AE) algorithm, not manual control of android.sensor.exposureTime and android.sensor.frameDuration.

android.control.aePrecaptureTrigger byte [public] [limited]
  • IDLE

    The trigger is idle.

  • START

    The precapture metering sequence will be started by the camera device.

    The exact effect of the precapture trigger depends on the current AE mode and state.

  • CANCEL

    The camera device will cancel any currently active or completed precapture metering sequence, the auto-exposure routine will return to its initial state.

Whether the camera device will trigger a precapture metering sequence when it processes this request.

Details

This entry is normally set to IDLE, or is not included at all in the request settings. When included and set to START, the camera device will trigger the auto-exposure (AE) precapture metering sequence.

When set to CANCEL, the camera device will cancel any active precapture metering trigger, and return to its initial AE state. If a precapture metering sequence is already completed, and the camera device has implicitly locked the AE for subsequent still capture, the CANCEL trigger will unlock the AE and return to its initial AE state.

The precapture sequence should be triggered before starting a high-quality still capture for final metering decisions to be made, and for firing pre-capture flash pulses to estimate scene brightness and required final capture flash power, when the flash is enabled.

Normally, this entry should be set to START for only a single request, and the application should wait until the sequence completes before starting a new one.

When a precapture metering sequence is finished, the camera device may lock the auto-exposure routine internally to be able to accurately expose the subsequent still capture image (android.control.captureIntent == STILL_CAPTURE). For this case, the AE may not resume normal scan if no subsequent still capture is submitted. To ensure that the AE routine restarts normal scan, the application should submit a request with android.control.aeLock == true, followed by a request with android.control.aeLock == false, if the application decides not to submit a still capture request after the precapture sequence completes. Alternatively, for API level 23 or newer devices, the CANCEL can be used to unlock the camera device internally locked AE if the application doesn't submit a still capture request after the AE precapture trigger. Note that, the CANCEL was added in API level 23, and must not be used in devices that have earlier API levels.

The exact effect of auto-exposure (AE) precapture trigger depends on the current AE mode and state; see android.control.aeState for AE precapture state transition details.

On LEGACY-level devices, the precapture trigger is not supported; capturing a high-resolution JPEG image will automatically trigger a precapture sequence before the high-resolution capture, including potentially firing a pre-capture flash.

Using the precapture trigger and the auto-focus trigger android.control.afTrigger simultaneously is allowed. However, since these triggers often require cooperation between the auto-focus and auto-exposure routines (for example, the may need to be enabled for a focus sweep), the camera device may delay acting on a later trigger until the previous trigger has been fully handled. This may lead to longer intervals between the trigger and changes to android.control.aeState indicating the start of the precapture sequence, for example.

If both the precapture and the auto-focus trigger are activated on the same request, then the camera device will complete them in the optimal order for that device.

HAL Implementation Details

The HAL must support triggering the AE precapture trigger while an AF trigger is active (and vice versa), or at the same time as the AF trigger. It is acceptable for the HAL to treat these as two consecutive triggers, for example handling the AF trigger and then the AE trigger. Or the HAL may choose to optimize the case with both triggers fired at once, to minimize the latency for converging both focus and exposure/flash usage.

android.control.afMode byte [public] [legacy]
  • OFF

    The auto-focus routine does not control the lens; android.lens.focusDistance is controlled by the application.

  • AUTO

    Basic automatic focus mode.

    In this mode, the lens does not move unless the autofocus trigger action is called. When that trigger is activated, AF will transition to ACTIVE_SCAN, then to the outcome of the scan (FOCUSED or NOT_FOCUSED).

    Always supported if lens is not fixed focus.

    Use android.lens.info.minimumFocusDistance to determine if lens is fixed-focus.

    Triggering AF_CANCEL resets the lens position to default, and sets the AF state to INACTIVE.

  • MACRO

    Close-up focusing mode.

    In this mode, the lens does not move unless the autofocus trigger action is called. When that trigger is activated, AF will transition to ACTIVE_SCAN, then to the outcome of the scan (FOCUSED or NOT_FOCUSED). This mode is optimized for focusing on objects very close to the camera.

    When that trigger is activated, AF will transition to ACTIVE_SCAN, then to the outcome of the scan (FOCUSED or NOT_FOCUSED). Triggering cancel AF resets the lens position to default, and sets the AF state to INACTIVE.

  • CONTINUOUS_VIDEO

    In this mode, the AF algorithm modifies the lens position continually to attempt to provide a constantly-in-focus image stream.

    The focusing behavior should be suitable for good quality video recording; typically this means slower focus movement and no overshoots. When the AF trigger is not involved, the AF algorithm should start in INACTIVE state, and then transition into PASSIVE_SCAN and PASSIVE_FOCUSED states as appropriate. When the AF trigger is activated, the algorithm should immediately transition into AF_FOCUSED or AF_NOT_FOCUSED as appropriate, and lock the lens position until a cancel AF trigger is received.

    Once cancel is received, the algorithm should transition back to INACTIVE and resume passive scan. Note that this behavior is not identical to CONTINUOUS_PICTURE, since an ongoing PASSIVE_SCAN must immediately be canceled.

  • CONTINUOUS_PICTURE

    In this mode, the AF algorithm modifies the lens position continually to attempt to provide a constantly-in-focus image stream.

    The focusing behavior should be suitable for still image capture; typically this means focusing as fast as possible. When the AF trigger is not involved, the AF algorithm should start in INACTIVE state, and then transition into PASSIVE_SCAN and PASSIVE_FOCUSED states as appropriate as it attempts to maintain focus. When the AF trigger is activated, the algorithm should finish its PASSIVE_SCAN if active, and then transition into AF_FOCUSED or AF_NOT_FOCUSED as appropriate, and lock the lens position until a cancel AF trigger is received.

    When the AF cancel trigger is activated, the algorithm should transition back to INACTIVE and then act as if it has just been started.

  • EDOF

    Extended depth of field (digital focus) mode.

    The camera device will produce images with an extended depth of field automatically; no special focusing operations need to be done before taking a picture.

    AF triggers are ignored, and the AF state will always be INACTIVE.

Whether auto-focus (AF) is currently enabled, and what mode it is set to.

android.control.afAvailableModes

Details

Only effective if android.control.mode = AUTO and the lens is not fixed focus (i.e. android.lens.info.minimumFocusDistance > 0). Also note that when android.control.aeMode is OFF, the behavior of AF is device dependent. It is recommended to lock AF by using android.control.afTrigger before setting android.control.aeMode to OFF, or set AF mode to OFF when AE is OFF.

If the lens is controlled by the camera device auto-focus algorithm, the camera device will report the current AF status in android.control.afState in result metadata.

HAL Implementation Details

When afMode is AUTO or MACRO, the lens must not move until an AF trigger is sent in a request (android.control.afTrigger == START). After an AF trigger, the afState will end up with either FOCUSED_LOCKED or NOT_FOCUSED_LOCKED state (see android.control.afState for detailed state transitions), which indicates that the lens is locked and will not move. If camera movement (e.g. tilting camera) causes the lens to move after the lens is locked, the HAL must compensate this movement appropriately such that the same focal plane remains in focus.

When afMode is one of the continuous auto focus modes, the HAL is free to start a AF scan whenever it's not locked. When the lens is locked after an AF trigger (see android.control.afState for detailed state transitions), the HAL should maintain the same lock behavior as above.

When afMode is OFF, the application controls focus manually. The accuracy of the focus distance control depends on the android.lens.info.focusDistanceCalibration. However, the lens must not move regardless of the camera movement for any focus distance manual control.

To put this in concrete terms, if the camera has lens elements which may move based on camera orientation or motion (e.g. due to gravity), then the HAL must drive the lens to remain in a fixed position invariant to the camera's orientation or motion, for example, by using accelerometer measurements in the lens control logic. This is a typical issue that will arise on camera modules with open-loop VCMs.

android.control.afRegions int32 x 5 x area_count [public as meteringRectangle]

List of metering areas to use for auto-focus.

Pixel coordinates within android.sensor.info.activeArraySize

Coordinates must be between [(0,0), (width, height)) of android.sensor.info.activeArraySize

Details

Not available if android.control.maxRegionsAf is 0. Otherwise will always be present.

The maximum number of focus areas supported by the device is determined by the value of android.control.maxRegionsAf.

The coordinate system is based on the active pixel array, with (0,0) being the top-left pixel in the active pixel array, and (android.sensor.info.activeArraySize.width - 1, android.sensor.info.activeArraySize.height - 1) being the bottom-right pixel in the active pixel array.

The weight must be within [0, 1000], and represents a weight for every pixel in the area. This means that a large metering area with the same weight as a smaller area will have more effect in the metering result. Metering areas can partially overlap and the camera device will add the weights in the overlap region.

The weights are relative to weights of other metering regions, so if only one region is used, all non-zero weights will have the same effect. A region with 0 weight is ignored.

If all regions have 0 weight, then no specific metering area needs to be used by the camera device.

If the metering region is outside the used android.scaler.cropRegion returned in capture result metadata, the camera device will ignore the sections outside the crop region and output only the intersection rectangle as the metering region in the result metadata. If the region is entirely outside the crop region, it will be ignored and not reported in the result metadata.

HAL Implementation Details

The HAL level representation of MeteringRectangle[] is a int[5 * area_count]. Every five elements represent a metering region of (xmin, ymin, xmax, ymax, weight). The rectangle is defined to be inclusive on xmin and ymin, but exclusive on xmax and ymax.

android.control.afTrigger byte [public] [legacy]
  • IDLE

    The trigger is idle.

  • START

    Autofocus will trigger now.

  • CANCEL

    Autofocus will return to its initial state, and cancel any currently active trigger.

Whether the camera device will trigger autofocus for this request.

Details

This entry is normally set to IDLE, or is not included at all in the request settings.

When included and set to START, the camera device will trigger the autofocus algorithm. If autofocus is disabled, this trigger has no effect.

When set to CANCEL, the camera device will cancel any active trigger, and return to its initial AF state.

Generally, applications should set this entry to START or CANCEL for only a single capture, and then return it to IDLE (or not set at all). Specifying START for multiple captures in a row means restarting the AF operation over and over again.

See android.control.afState for what the trigger means for each AF mode.

Using the autofocus trigger and the precapture trigger android.control.aePrecaptureTrigger simultaneously is allowed. However, since these triggers often require cooperation between the auto-focus and auto-exposure routines (for example, the may need to be enabled for a focus sweep), the camera device may delay acting on a later trigger until the previous trigger has been fully handled. This may lead to longer intervals between the trigger and changes to android.control.afState, for example.

HAL Implementation Details

The HAL must support triggering the AF trigger while an AE precapture trigger is active (and vice versa), or at the same time as the AE trigger. It is acceptable for the HAL to treat these as two consecutive triggers, for example handling the AF trigger and then the AE trigger. Or the HAL may choose to optimize the case with both triggers fired at once, to minimize the latency for converging both focus and exposure/flash usage.

android.control.awbLock byte [public as boolean] [legacy]
  • OFF

    Auto-white balance lock is disabled; the AWB algorithm is free to update its parameters if in AUTO mode.

  • ON

    Auto-white balance lock is enabled; the AWB algorithm will not update its parameters while the lock is active.

Whether auto-white balance (AWB) is currently locked to its latest calculated values.

Details

When set to true (ON), the AWB algorithm is locked to its latest parameters, and will not change color balance settings until the lock is set to false (OFF).

Since the camera device has a pipeline of in-flight requests, the settings that get locked do not necessarily correspond to the settings that were present in the latest capture result received from the camera device, since additional captures and AWB updates may have occurred even before the result was sent out. If an application is switching between automatic and manual control and wishes to eliminate any flicker during the switch, the following procedure is recommended:

  1. Starting in auto-AWB mode:
  2. Lock AWB
  3. Wait for the first result to be output that has the AWB locked
  4. Copy AWB settings from that result into a request, set the request to manual AWB
  5. Submit the capture request, proceed to run manual AWB as desired.

Note that AWB lock is only meaningful when android.control.awbMode is in the AUTO mode; in other modes, AWB is already fixed to a specific setting.

Some LEGACY devices may not support ON; the value is then overridden to OFF.

android.control.awbMode byte [public] [legacy]
  • OFF

    The camera device's auto-white balance routine is disabled.

    The application-selected color transform matrix (android.colorCorrection.transform) and gains (android.colorCorrection.gains) are used by the camera device for manual white balance control.

  • AUTO

    The camera device's auto-white balance routine is active.

    The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request.

  • INCANDESCENT

    The camera device's auto-white balance routine is disabled; the camera device uses incandescent light as the assumed scene illumination for white balance.

    While the exact white balance transforms are up to the camera device, they will approximately match the CIE standard illuminant A.

    The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request.

  • FLUORESCENT

    The camera device's auto-white balance routine is disabled; the camera device uses fluorescent light as the assumed scene illumination for white balance.

    While the exact white balance transforms are up to the camera device, they will approximately match the CIE standard illuminant F2.

    The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request.

  • WARM_FLUORESCENT

    The camera device's auto-white balance routine is disabled; the camera device uses warm fluorescent light as the assumed scene illumination for white balance.

    While the exact white balance transforms are up to the camera device, they will approximately match the CIE standard illuminant F4.

    The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request.

  • DAYLIGHT

    The camera device's auto-white balance routine is disabled; the camera device uses daylight light as the assumed scene illumination for white balance.

    While the exact white balance transforms are up to the camera device, they will approximately match the CIE standard illuminant D65.

    The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request.

  • CLOUDY_DAYLIGHT

    The camera device's auto-white balance routine is disabled; the camera device uses cloudy daylight light as the assumed scene illumination for white balance.

    The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request.

  • TWILIGHT

    The camera device's auto-white balance routine is disabled; the camera device uses twilight light as the assumed scene illumination for white balance.

    The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request.

  • SHADE

    The camera device's auto-white balance routine is disabled; the camera device uses shade light as the assumed scene illumination for white balance.

    The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request.

Whether auto-white balance (AWB) is currently setting the color transform fields, and what its illumination target is.

android.control.awbAvailableModes

Details

This control is only effective if android.control.mode is AUTO.

When set to the ON mode, the camera device's auto-white balance routine is enabled, overriding the application's selected android.colorCorrection.transform, android.colorCorrection.gains and android.colorCorrection.mode. Note that when android.control.aeMode is OFF, the behavior of AWB is device dependent. It is recommened to also set AWB mode to OFF or lock AWB by using android.control.awbLock before setting AE mode to OFF.

When set to the OFF mode, the camera device's auto-white balance routine is disabled. The application manually controls the white balance by android.colorCorrection.transform, android.colorCorrection.gains and android.colorCorrection.mode.

When set to any other modes, the camera device's auto-white balance routine is disabled. The camera device uses each particular illumination target for white balance adjustment. The application's values for android.colorCorrection.transform, android.colorCorrection.gains and android.colorCorrection.mode are ignored.

android.control.awbRegions int32 x 5 x area_count [public as meteringRectangle]

List of metering areas to use for auto-white-balance illuminant estimation.

Pixel coordinates within android.sensor.info.activeArraySize

Coordinates must be between [(0,0), (width, height)) of android.sensor.info.activeArraySize

Details

Not available if android.control.maxRegionsAwb is 0. Otherwise will always be present.

The maximum number of regions supported by the device is determined by the value of android.control.maxRegionsAwb.

The coordinate system is based on the active pixel array, with (0,0) being the top-left pixel in the active pixel array, and (android.sensor.info.activeArraySize.width - 1, android.sensor.info.activeArraySize.height - 1) being the bottom-right pixel in the active pixel array.

The weight must range from 0 to 1000, and represents a weight for every pixel in the area. This means that a large metering area with the same weight as a smaller area will have more effect in the metering result. Metering areas can partially overlap and the camera device will add the weights in the overlap region.

The weights are relative to weights of other white balance metering regions, so if only one region is used, all non-zero weights will have the same effect. A region with 0 weight is ignored.

If all regions have 0 weight, then no specific metering area needs to be used by the camera device.

If the metering region is outside the used android.scaler.cropRegion returned in capture result metadata, the camera device will ignore the sections outside the crop region and output only the intersection rectangle as the metering region in the result metadata. If the region is entirely outside the crop region, it will be ignored and not reported in the result metadata.

HAL Implementation Details

The HAL level representation of MeteringRectangle[] is a int[5 * area_count]. Every five elements represent a metering region of (xmin, ymin, xmax, ymax, weight). The rectangle is defined to be inclusive on xmin and ymin, but exclusive on xmax and ymax.

android.control.captureIntent byte [public] [legacy]
  • CUSTOM

    The goal of this request doesn't fall into the other categories. The camera device will default to preview-like behavior.

  • PREVIEW

    This request is for a preview-like use case.

    The precapture trigger may be used to start off a metering w/flash sequence.

  • STILL_CAPTURE

    This request is for a still capture-type use case.

    If the flash unit is under automatic control, it may fire as needed.

  • VIDEO_RECORD

    This request is for a video recording use case.

  • VIDEO_SNAPSHOT

    This request is for a video snapshot (still image while recording video) use case.

    The camera device should take the highest-quality image possible (given the other settings) without disrupting the frame rate of video recording.

  • ZERO_SHUTTER_LAG

    This request is for a ZSL usecase; the application will stream full-resolution images and reprocess one or several later for a final capture.

  • MANUAL

    This request is for manual capture use case where the applications want to directly control the capture parameters.

    For example, the application may wish to manually control android.sensor.exposureTime, android.sensor.sensitivity, etc.

Information to the camera device 3A (auto-exposure, auto-focus, auto-white balance) routines about the purpose of this capture, to help the camera device to decide optimal 3A strategy.

Details

This control (except for MANUAL) is only effective if android.control.mode != OFF and any 3A routine is active.

ZERO_SHUTTER_LAG will be supported if android.request.availableCapabilities contains PRIVATE_REPROCESSING or YUV_REPROCESSING. MANUAL will be supported if android.request.availableCapabilities contains MANUAL_SENSOR. Other intent values are always supported.

android.control.effectMode byte [public] [legacy]
  • OFF

    No color effect will be applied.

  • MONO [optional]

    A "monocolor" effect where the image is mapped into a single color.

    This will typically be grayscale.

  • NEGATIVE [optional]

    A "photo-negative" effect where the image's colors are inverted.

  • SOLARIZE [optional]

    A "solarisation" effect (Sabattier effect) where the image is wholly or partially reversed in tone.

  • SEPIA [optional]

    A "sepia" effect where the image is mapped into warm gray, red, and brown tones.

  • POSTERIZE [optional]

    A "posterization" effect where the image uses discrete regions of tone rather than a continuous gradient of tones.

  • WHITEBOARD [optional]

    A "whiteboard" effect where the image is typically displayed as regions of white, with black or grey details.

  • BLACKBOARD [optional]

    A "blackboard" effect where the image is typically displayed as regions of black, with white or grey details.

  • AQUA [optional]

    An "aqua" effect where a blue hue is added to the image.

A special color effect to apply.

android.control.availableEffects

Details

When this mode is set, a color effect will be applied to images produced by the camera device. The interpretation and implementation of these color effects is left to the implementor of the camera device, and should not be depended on to be consistent (or present) across all devices.

android.control.mode byte [public] [legacy]
  • OFF

    Full application control of pipeline.

    All control by the device's metering and focusing (3A) routines is disabled, and no other settings in android.control.* have any effect, except that android.control.captureIntent may be used by the camera device to select post-processing values for processing blocks that do not allow for manual control, or are not exposed by the camera API.

    However, the camera device's 3A routines may continue to collect statistics and update their internal state so that when control is switched to AUTO mode, good control values can be immediately applied.

  • AUTO

    Use settings for each individual 3A routine.

    Manual control of capture parameters is disabled. All controls in android.control.* besides sceneMode take effect.

  • USE_SCENE_MODE [optional]

    Use a specific scene mode.

    Enabling this disables control.aeMode, control.awbMode and control.afMode controls; the camera device will ignore those settings while USE_SCENE_MODE is active (except for FACE_PRIORITY scene mode). Other control entries are still active. This setting can only be used if scene mode is supported (i.e. android.control.availableSceneModes contain some modes other than DISABLED).

  • OFF_KEEP_STATE [optional]

    Same as OFF mode, except that this capture will not be used by camera device background auto-exposure, auto-white balance and auto-focus algorithms (3A) to update their statistics.

    Specifically, the 3A routines are locked to the last values set from a request with AUTO, OFF, or USE_SCENE_MODE, and any statistics or state updates collected from manual captures with OFF_KEEP_STATE will be discarded by the camera device.

Overall mode of 3A (auto-exposure, auto-white-balance, auto-focus) control routines.

android.control.availableModes

Details

This is a top-level 3A control switch. When set to OFF, all 3A control by the camera device is disabled. The application must set the fields for capture parameters itself.

When set to AUTO, the individual algorithm controls in android.control.* are in effect, such as android.control.afMode.

When set to USE_SCENE_MODE, the individual controls in android.control.* are mostly disabled, and the camera device implements one of the scene mode settings (such as ACTION, SUNSET, or PARTY) as it wishes. The camera device scene mode 3A settings are provided by capture results.

When set to OFF_KEEP_STATE, it is similar to OFF mode, the only difference is that this frame will not be used by camera device background 3A statistics update, as if this frame is never captured. This mode can be used in the scenario where the application doesn't want a 3A manual control capture to affect the subsequent auto 3A capture results.

android.control.sceneMode byte [public] [legacy]
  • DISABLED 0

    Indicates that no scene modes are set for a given capture request.

  • FACE_PRIORITY

    If face detection support exists, use face detection data for auto-focus, auto-white balance, and auto-exposure routines.

    If face detection statistics are disabled (i.e. android.statistics.faceDetectMode is set to OFF), this should still operate correctly (but will not return face detection statistics to the framework).

    Unlike the other scene modes, android.control.aeMode, android.control.awbMode, and android.control.afMode remain active when FACE_PRIORITY is set.

  • ACTION [optional]

    Optimized for photos of quickly moving objects.

    Similar to SPORTS.

  • PORTRAIT [optional]

    Optimized for still photos of people.

  • LANDSCAPE [optional]

    Optimized for photos of distant macroscopic objects.

  • NIGHT [optional]

    Optimized for low-light settings.

  • NIGHT_PORTRAIT [optional]

    Optimized for still photos of people in low-light settings.

  • THEATRE [optional]

    Optimized for dim, indoor settings where flash must remain off.

  • BEACH [optional]

    Optimized for bright, outdoor beach settings.

  • SNOW [optional]

    Optimized for bright, outdoor settings containing snow.

  • SUNSET [optional]

    Optimized for scenes of the setting sun.

  • STEADYPHOTO [optional]

    Optimized to avoid blurry photos due to small amounts of device motion (for example: due to hand shake).

  • FIREWORKS [optional]

    Optimized for nighttime photos of fireworks.

  • SPORTS [optional]

    Optimized for photos of quickly moving people.

    Similar to ACTION.

  • PARTY [optional]

    Optimized for dim, indoor settings with multiple moving people.

  • CANDLELIGHT [optional]

    Optimized for dim settings where the main light source is a flame.

  • BARCODE [optional]

    Optimized for accurately capturing a photo of barcode for use by camera applications that wish to read the barcode value.

  • HIGH_SPEED_VIDEO [deprecated] [optional]

    This is deprecated, please use CameraDevice#createConstrainedHighSpeedCaptureSession and CameraConstrainedHighSpeedCaptureSession#createHighSpeedRequestList for high speed video recording.

    Optimized for high speed video recording (frame rate >=60fps) use case.

    The supported high speed video sizes and fps ranges are specified in android.control.availableHighSpeedVideoConfigurations. To get desired output frame rates, the application is only allowed to select video size and fps range combinations listed in this static metadata. The fps range can be control via android.control.aeTargetFpsRange.

    In this mode, the camera device will override aeMode, awbMode, and afMode to ON, ON, and CONTINUOUS_VIDEO, respectively. All post-processing block mode controls will be overridden to be FAST. Therefore, no manual control of capture and post-processing parameters is possible. All other controls operate the same as when android.control.mode == AUTO. This means that all other android.control.* fields continue to work, such as

    Outside of android.control.*, the following controls will work:

    For high speed recording use case, the actual maximum supported frame rate may be lower than what camera can output, depending on the destination Surfaces for the image data. For example, if the destination surface is from video encoder, the application need check if the video encoder is capable of supporting the high frame rate for a given video size, or it will end up with lower recording frame rate. If the destination surface is from preview window, the preview frame rate will be bounded by the screen refresh rate.

    The camera device will only support up to 2 output high speed streams (processed non-stalling format defined in android.request.maxNumOutputStreams) in this mode. This control will be effective only if all of below conditions are true:

    When above conditions are NOT satistied, the controls of this mode and android.control.aeTargetFpsRange will be ignored by the camera device, the camera device will fall back to android.control.mode == AUTO, and the returned capture result metadata will give the fps range choosen by the camera device.

    Switching into or out of this mode may trigger some camera ISP/sensor reconfigurations, which may introduce extra latency. It is recommended that the application avoids unnecessary scene mode switch as much as possible.

  • HDR [optional]

    Turn on a device-specific high dynamic range (HDR) mode.

    In this scene mode, the camera device captures images that keep a larger range of scene illumination levels visible in the final image. For example, when taking a picture of a object in front of a bright window, both the object and the scene through the window may be visible when using HDR mode, while in normal AUTO mode, one or the other may be poorly exposed. As a tradeoff, HDR mode generally takes much longer to capture a single image, has no user control, and may have other artifacts depending on the HDR method used.

    Therefore, HDR captures operate at a much slower rate than regular captures.

    In this mode, on LIMITED or FULL devices, when a request is made with a android.control.captureIntent of STILL_CAPTURE, the camera device will capture an image using a high dynamic range capture technique. On LEGACY devices, captures that target a JPEG-format output will be captured with HDR, and the capture intent is not relevant.

    The HDR capture may involve the device capturing a burst of images internally and combining them into one, or it may involve the device using specialized high dynamic range capture hardware. In all cases, a single image is produced in response to a capture request submitted while in HDR mode.

    Since substantial post-processing is generally needed to produce an HDR image, only YUV and JPEG outputs are supported for LIMITED/FULL device HDR captures, and only JPEG outputs are supported for LEGACY HDR captures. Using a RAW output for HDR capture is not supported.

  • FACE_PRIORITY_LOW_LIGHT [optional] [hidden]

    Same as FACE_PRIORITY scene mode, except that the camera device will choose higher sensitivity values (android.sensor.sensitivity) under low light conditions.

    The camera device may be tuned to expose the images in a reduced sensitivity range to produce the best quality images. For example, if the android.sensor.info.sensitivityRange gives range of [100, 1600], the camera device auto-exposure routine tuning process may limit the actual exposure sensitivity range to [100, 1200] to ensure that the noise level isn't exessive in order to preserve the image quality. Under this situation, the image under low light may be under-exposed when the sensor max exposure time (bounded by the android.control.aeTargetFpsRange when android.control.aeMode is one of the ON_* modes) and effective max sensitivity are reached. This scene mode allows the camera device auto-exposure routine to increase the sensitivity up to the max sensitivity specified by android.sensor.info.sensitivityRange when the scene is too dark and the max exposure time is reached. The captured images may be noisier compared with the images captured in normal FACE_PRIORITY mode; therefore, it is recommended that the application only use this scene mode when it is capable of reducing the noise level of the captured images.

    Unlike the other scene modes, android.control.aeMode, android.control.awbMode, and android.control.afMode remain active when FACE_PRIORITY_LOW_LIGHT is set.

Control for which scene mode is currently active.

android.control.availableSceneModes

Details

Scene modes are custom camera modes optimized for a certain set of conditions and capture settings.

This is the mode that that is active when android.control.mode == USE_SCENE_MODE. Aside from FACE_PRIORITY, these modes will disable android.control.aeMode, android.control.awbMode, and android.control.afMode while in use.

The interpretation and implementation of these scene modes is left to the implementor of the camera device. Their behavior will not be consistent across all devices, and any given device may only implement a subset of these modes.

HAL Implementation Details

HAL implementations that include scene modes are expected to provide the per-scene settings to use for android.control.aeMode, android.control.awbMode, and android.control.afMode in android.control.sceneModeOverrides.

For HIGH_SPEED_VIDEO mode, if it is included in android.control.availableSceneModes, the HAL must list supported video size and fps range in android.control.availableHighSpeedVideoConfigurations. For a given size, e.g. 1280x720, if the HAL has two different sensor configurations for normal streaming mode and high speed streaming, when this scene mode is set/reset in a sequence of capture requests, the HAL may have to switch between different sensor modes. This mode is deprecated in HAL3.3, to support high speed video recording, please implement android.control.availableHighSpeedVideoConfigurations and CONSTRAINED_HIGH_SPEED_VIDEO capbility defined in android.request.availableCapabilities.

android.control.videoStabilizationMode byte [public] [legacy]
  • OFF

    Video stabilization is disabled.

  • ON

    Video stabilization is enabled.

Whether video stabilization is active.

Details

Video stabilization automatically warps images from the camera in order to stabilize motion between consecutive frames.

If enabled, video stabilization can modify the android.scaler.cropRegion to keep the video stream stabilized.

Switching between different video stabilization modes may take several frames to initialize, the camera device will report the current mode in capture result metadata. For example, When "ON" mode is requested, the video stabilization modes in the first several capture results may still be "OFF", and it will become "ON" when the initialization is done.

In addition, not all recording sizes or frame rates may be supported for stabilization by a device that reports stabilization support. It is guaranteed that an output targeting a MediaRecorder or MediaCodec will be stabilized if the recording resolution is less than or equal to 1920 x 1080 (width less than or equal to 1920, height less than or equal to 1080), and the recording frame rate is less than or equal to 30fps. At other sizes, the CaptureResult android.control.videoStabilizationMode field will return OFF if the recording output is not stabilized, or if there are no output Surface types that can be stabilized.

If a camera device supports both this mode and OIS (android.lens.opticalStabilizationMode), turning both modes on may produce undesirable interaction, so it is recommended not to enable both at the same time.

static
Property Name Type Description Units Range Tags
android.control.aeAvailableAntibandingModes byte x n [public as enumList] [legacy]
list of enums

List of auto-exposure antibanding modes for android.control.aeAntibandingMode that are supported by this camera device.

Any value listed in android.control.aeAntibandingMode

Details

Not all of the auto-exposure anti-banding modes may be supported by a given camera device. This field lists the valid anti-banding modes that the application may request for this camera device with the android.control.aeAntibandingMode control.

android.control.aeAvailableModes byte x n [public as enumList] [legacy]
list of enums

List of auto-exposure modes for android.control.aeMode that are supported by this camera device.

Any value listed in android.control.aeMode

Details

Not all the auto-exposure modes may be supported by a given camera device, especially if no flash unit is available. This entry lists the valid modes for android.control.aeMode for this camera device.

All camera devices support ON, and all camera devices with flash units support ON_AUTO_FLASH and ON_ALWAYS_FLASH.

FULL mode camera devices always support OFF mode, which enables application control of camera exposure time, sensitivity, and frame duration.

LEGACY mode camera devices never support OFF mode. LIMITED mode devices support OFF if they support the MANUAL_SENSOR capability.

android.control.aeAvailableTargetFpsRanges int32 x 2 x n [public as rangeInt] [legacy]
list of pairs of frame rates

List of frame rate ranges for android.control.aeTargetFpsRange supported by this camera device.

Frames per second (FPS)
Details

For devices at the LEGACY level or above:

  • This list will always include (30, 30).
  • Also, for constant-framerate recording, for each normal CamcorderProfile that has quality in the range [QUALITY_LOW, QUALITY_2160P], if the profile is supported by the device and has videoFrameRate x, this list will always include (x,x).
  • For preview streaming use case, this list will always include (min, max) where min <= 15 and max >= 30.

For devices at the LIMITED level or above:

  • For YUV_420_888 burst capture use case, this list will always include (min, max) and (max, max) where min <= 15 and max = the maximum output frame rate of the maximum YUV_420_888 output size.
android.control.aeCompensationRange int32 x 2 [public as rangeInt] [legacy]

Maximum and minimum exposure compensation values for android.control.aeExposureCompensation, in counts of android.control.aeCompensationStep, that are supported by this camera device.

Range [0,0] indicates that exposure compensation is not supported.

For LIMITED and FULL devices, range must follow below requirements if exposure compensation is supported (range != [0, 0]):

Min.exposure compensation * android.control.aeCompensationStep <= -2 EV

Max.exposure compensation * android.control.aeCompensationStep >= 2 EV

LEGACY devices may support a smaller range than this.

android.control.aeCompensationStep rational [public] [legacy]

Smallest step by which the exposure compensation can be changed.

Exposure Value (EV)
Details

This is the unit for android.control.aeExposureCompensation. For example, if this key has a value of 1/2, then a setting of -2 for android.control.aeExposureCompensation means that the target EV offset for the auto-exposure routine is -1 EV.

One unit of EV compensation changes the brightness of the captured image by a factor of two. +1 EV doubles the image brightness, while -1 EV halves the image brightness.

HAL Implementation Details

This must be less than or equal to 1/2.

android.control.afAvailableModes byte x n [public as enumList] [legacy]
List of enums

List of auto-focus (AF) modes for android.control.afMode that are supported by this camera device.

Any value listed in android.control.afMode

Details

Not all the auto-focus modes may be supported by a given camera device. This entry lists the valid modes for android.control.afMode for this camera device.

All LIMITED and FULL mode camera devices will support OFF mode, and all camera devices with adjustable focuser units (android.lens.info.minimumFocusDistance > 0) will support AUTO mode.

LEGACY devices will support OFF mode only if they support focusing to infinity (by also setting android.lens.focusDistance to 0.0f).

android.control.availableEffects byte x n [public as enumList] [legacy]
List of enums (android.control.effectMode).

List of color effects for android.control.effectMode that are supported by this camera device.

Any value listed in android.control.effectMode

Details

This list contains the color effect modes that can be applied to images produced by the camera device. Implementations are not expected to be consistent across all devices. If no color effect modes are available for a device, this will only list OFF.

A color effect will only be applied if android.control.mode != OFF. OFF is always included in this list.

This control has no effect on the operation of other control routines such as auto-exposure, white balance, or focus.

android.control.availableSceneModes byte x n [public as enumList] [legacy]
List of enums (android.control.sceneMode).

List of scene modes for android.control.sceneMode that are supported by this camera device.

Any value listed in android.control.sceneMode

Details

This list contains scene modes that can be set for the camera device. Only scene modes that have been fully implemented for the camera device may be included here. Implementations are not expected to be consistent across all devices.

If no scene modes are supported by the camera device, this will be set to DISABLED. Otherwise DISABLED will not be listed.

FACE_PRIORITY is always listed if face detection is supported (i.e.android.statistics.info.maxFaceCount > 0).

android.control.availableVideoStabilizationModes byte x n [public as enumList] [legacy]
List of enums.

List of video stabilization modes for android.control.videoStabilizationMode that are supported by this camera device.

Any value listed in android.control.videoStabilizationMode

Details

OFF will always be listed.

android.control.awbAvailableModes byte x n [public as enumList] [legacy]
List of enums

List of auto-white-balance modes for android.control.awbMode that are supported by this camera device.

Any value listed in android.control.awbMode

Details

Not all the auto-white-balance modes may be supported by a given camera device. This entry lists the valid modes for android.control.awbMode for this camera device.

All camera devices will support ON mode.

Camera devices that support the MANUAL_POST_PROCESSING capability will always support OFF mode, which enables application control of white balance, by using android.colorCorrection.transform and android.colorCorrection.gains(android.colorCorrection.mode must be set to TRANSFORM_MATRIX). This includes all FULL mode camera devices.

android.control.maxRegions int32 x 3 [hidden] [legacy]

List of the maximum number of regions that can be used for metering in auto-exposure (AE), auto-white balance (AWB), and auto-focus (AF); this corresponds to the the maximum number of elements in android.control.aeRegions, android.control.awbRegions, and android.control.afRegions.

Value must be >= 0 for each element. For full-capability devices this value must be >= 1 for AE and AF. The order of the elements is: (AE, AWB, AF).

android.control.maxRegionsAe int32 [public] [synthetic] [legacy]

The maximum number of metering regions that can be used by the auto-exposure (AE) routine.

Value will be >= 0. For FULL-capability devices, this value will be >= 1.

Details

This corresponds to the the maximum allowed number of elements in android.control.aeRegions.

HAL Implementation Details

This entry is private to the framework. Fill in maxRegions to have this entry be automatically populated.

android.control.maxRegionsAwb int32 [public] [synthetic] [legacy]

The maximum number of metering regions that can be used by the auto-white balance (AWB) routine.

Value will be >= 0.

Details

This corresponds to the the maximum allowed number of elements in android.control.awbRegions.

HAL Implementation Details

This entry is private to the framework. Fill in maxRegions to have this entry be automatically populated.

android.control.maxRegionsAf int32 [public] [synthetic] [legacy]

The maximum number of metering regions that can be used by the auto-focus (AF) routine.

Value will be >= 0. For FULL-capability devices, this value will be >= 1.

Details

This corresponds to the the maximum allowed number of elements in android.control.afRegions.

HAL Implementation Details

This entry is private to the framework. Fill in maxRegions to have this entry be automatically populated.

android.control.sceneModeOverrides byte x 3 x length(availableSceneModes) [system] [limited]

Ordered list of auto-exposure, auto-white balance, and auto-focus settings to use with each available scene mode.

For each available scene mode, the list must contain three entries containing the android.control.aeMode, android.control.awbMode, and android.control.afMode values used by the camera device. The entry order is (aeMode, awbMode, afMode) where aeMode has the lowest index position.

Details

When a scene mode is enabled, the camera device is expected to override android.control.aeMode, android.control.awbMode, and android.control.afMode with its preferred settings for that scene mode.

The order of this list matches that of availableSceneModes, with 3 entries for each mode. The overrides listed for FACE_PRIORITY and FACE_PRIORITY_LOW_LIGHT (if supported) are ignored, since for that mode the application-set android.control.aeMode, android.control.awbMode, and android.control.afMode values are used instead, matching the behavior when android.control.mode is set to AUTO. It is recommended that the FACE_PRIORITY and FACE_PRIORITY_LOW_LIGHT (if supported) overrides should be set to 0.

For example, if availableSceneModes contains (FACE_PRIORITY, ACTION, NIGHT), then the camera framework expects sceneModeOverrides to have 9 entries formatted like: (0, 0, 0, ON_AUTO_FLASH, AUTO, CONTINUOUS_PICTURE, ON_AUTO_FLASH, INCANDESCENT, AUTO).

HAL Implementation Details

To maintain backward compatibility, this list will be made available in the static metadata of the camera service. The camera service will use these values to set android.control.aeMode, android.control.awbMode, and android.control.afMode when using a scene mode other than FACE_PRIORITY and FACE_PRIORITY_LOW_LIGHT (if supported).

android.control.availableHighSpeedVideoConfigurations int32 x 5 x n [hidden as highSpeedVideoConfiguration] [limited]

List of available high speed video size, fps range and max batch size configurations supported by the camera device, in the format of (width, height, fps_min, fps_max, batch_size_max).

For each configuration, the fps_max >= 120fps.

Details

When CONSTRAINED_HIGH_SPEED_VIDEO is supported in android.request.availableCapabilities, this metadata will list the supported high speed video size, fps range and max batch size configurations. All the sizes listed in this configuration will be a subset of the sizes reported by StreamConfigurationMap#getOutputSizes for processed non-stalling formats.

For the high speed video use case, the application must select the video size and fps range from this metadata to configure the recording and preview streams and setup the recording requests. For example, if the application intends to do high speed recording, it can select the maximum size reported by this metadata to configure output streams. Once the size is selected, application can filter this metadata by selected size and get the supported fps ranges, and use these fps ranges to setup the recording requests. Note that for the use case of multiple output streams, application must select one unique size from this metadata to use (e.g., preview and recording streams must have the same size). Otherwise, the high speed capture session creation will fail.

The min and max fps will be multiple times of 30fps.

High speed video streaming extends significant performance pressue to camera hardware, to achieve efficient high speed streaming, the camera device may have to aggregate multiple frames together and send to camera device for processing where the request controls are same for all the frames in this batch. Max batch size indicates the max possible number of frames the camera device will group together for this high speed stream configuration. This max batch size will be used to generate a high speed recording request list by CameraConstrainedHighSpeedCaptureSession#createHighSpeedRequestList. The max batch size for each configuration will satisfy below conditions:

  • Each max batch size will be a divisor of its corresponding fps_max / 30. For example, if max_fps is 300, max batch size will only be 1, 2, 5, or 10.
  • The camera device may choose smaller internal batch size for each configuration, but the actual batch size will be a divisor of max batch size. For example, if the max batch size is 8, the actual batch size used by camera device will only be 1, 2, 4, or 8.
  • The max batch size in each configuration entry must be no larger than 32.

The camera device doesn't have to support batch mode to achieve high speed video recording, in such case, batch_size_max will be reported as 1 in each configuration entry.

This fps ranges in this configuration list can only be used to create requests that are submitted to a high speed camera capture session created by CameraDevice#createConstrainedHighSpeedCaptureSession. The fps ranges reported in this metadata must not be used to setup capture requests for normal capture session, or it will cause request error.

HAL Implementation Details

All the sizes listed in this configuration will be a subset of the sizes reported by android.scaler.availableStreamConfigurations for processed non-stalling output formats. Note that for all high speed video configurations, HAL must be able to support a minimum of two streams, though the application might choose to configure just one stream.

The HAL may support multiple sensor modes for high speed outputs, for example, 120fps sensor mode and 120fps recording, 240fps sensor mode for 240fps recording. The application usually starts preview first, then starts recording. To avoid sensor mode switch caused stutter when starting recording as much as possible, the application may want to ensure the same sensor mode is used for preview and recording. Therefore, The HAL must advertise the variable fps range [30, fps_max] for each fixed fps range in this configuration list. For example, if the HAL advertises [120, 120] and [240, 240], the HAL must also advertise [30, 120] and [30, 240] for each configuration. In doing so, if the application intends to do 120fps recording, it can select [30, 120] to start preview, and [120, 120] to start recording. For these variable fps ranges, it's up to the HAL to decide the actual fps values that are suitable for smooth preview streaming. If the HAL sees different max_fps values that fall into different sensor modes in a sequence of requests, the HAL must switch the sensor mode as quick as possible to minimize the mode switch caused stutter.

android.control.aeLockAvailable byte [public as boolean] [legacy]
  • FALSE
  • TRUE

Whether the camera device supports android.control.aeLock

Details

Devices with MANUAL_SENSOR capability or BURST_CAPTURE capability will always list true. This includes FULL devices.

android.control.awbLockAvailable byte [public as boolean] [legacy]
  • FALSE
  • TRUE

Whether the camera device supports android.control.awbLock

Details

Devices with MANUAL_POST_PROCESSING capability or BURST_CAPTURE capability will always list true. This includes FULL devices.

android.control.availableModes byte x n [public as enumList] [legacy]
List of enums (android.control.mode).

List of control modes for android.control.mode that are supported by this camera device.

Any value listed in android.control.mode

Details

This list contains control modes that can be set for the camera device. LEGACY mode devices will always support AUTO mode. LIMITED and FULL devices will always support OFF, AUTO modes.

dynamic
Property Name Type Description Units Range Tags
android.control.aePrecaptureId int32 [system] [deprecated]

The ID sent with the latest CAMERA2_TRIGGER_PRECAPTURE_METERING call

Deprecated. Do not use.

Details

Must be 0 if no CAMERA2_TRIGGER_PRECAPTURE_METERING trigger received yet by HAL. Always updated even if AE algorithm ignores the trigger

android.control.aeAntibandingMode byte [public] [legacy]
  • OFF

    The camera device will not adjust exposure duration to avoid banding problems.

  • 50HZ

    The camera device will adjust exposure duration to avoid banding problems with 50Hz illumination sources.

  • 60HZ

    The camera device will adjust exposure duration to avoid banding problems with 60Hz illumination sources.

  • AUTO

    The camera device will automatically adapt its antibanding routine to the current illumination condition. This is the default mode if AUTO is available on given camera device.

The desired setting for the camera device's auto-exposure algorithm's antibanding compensation.

android.control.aeAvailableAntibandingModes

Details

Some kinds of lighting fixtures, such as some fluorescent lights, flicker at the rate of the power supply frequency (60Hz or 50Hz, depending on country). While this is typically not noticeable to a person, it can be visible to a camera device. If a camera sets its exposure time to the wrong value, the flicker may become visible in the viewfinder as flicker or in a final captured image, as a set of variable-brightness bands across the image.

Therefore, the auto-exposure routines of camera devices include antibanding routines that ensure that the chosen exposure value will not cause such banding. The choice of exposure time depends on the rate of flicker, which the camera device can detect automatically, or the expected rate can be selected by the application using this control.

A given camera device may not support all of the possible options for the antibanding mode. The android.control.aeAvailableAntibandingModes key contains the available modes for a given camera device.

AUTO mode is the default if it is available on given camera device. When AUTO mode is not available, the default will be either 50HZ or 60HZ, and both 50HZ and 60HZ will be available.

If manual exposure control is enabled (by setting android.control.aeMode or android.control.mode to OFF), then this setting has no effect, and the application must ensure it selects exposure times that do not cause banding issues. The android.statistics.sceneFlicker key can assist the application in this.

HAL Implementation Details

For all capture request templates, this field must be set to AUTO if AUTO mode is available. If AUTO is not available, the default must be either 50HZ or 60HZ, and both 50HZ and 60HZ must be available.

If manual exposure control is enabled (by setting android.control.aeMode or android.control.mode to OFF), then the exposure values provided by the application must not be adjusted for antibanding.

android.control.aeExposureCompensation int32 [public] [legacy]

Adjustment to auto-exposure (AE) target image brightness.

Compensation steps

android.control.aeCompensationRange

Details

The adjustment is measured as a count of steps, with the step size defined by android.control.aeCompensationStep and the allowed range by android.control.aeCompensationRange.

For example, if the exposure value (EV) step is 0.333, '6' will mean an exposure compensation of +2 EV; -3 will mean an exposure compensation of -1 EV. One EV represents a doubling of image brightness. Note that this control will only be effective if android.control.aeMode != OFF. This control will take effect even when android.control.aeLock == true.

In the event of exposure compensation value being changed, camera device may take several frames to reach the newly requested exposure target. During that time, android.control.aeState field will be in the SEARCHING state. Once the new exposure target is reached, android.control.aeState will change from SEARCHING to either CONVERGED, LOCKED (if AE lock is enabled), or FLASH_REQUIRED (if the scene is too dark for still capture).

android.control.aeLock byte [public as boolean] [legacy]
  • OFF

    Auto-exposure lock is disabled; the AE algorithm is free to update its parameters.

  • ON

    Auto-exposure lock is enabled; the AE algorithm must not update the exposure and sensitivity parameters while the lock is active.

    android.control.aeExposureCompensation setting changes will still take effect while auto-exposure is locked.

    Some rare LEGACY devices may not support this, in which case the value will always be overridden to OFF.

Whether auto-exposure (AE) is currently locked to its latest calculated values.

Details

When set to true (ON), the AE algorithm is locked to its latest parameters, and will not change exposure settings until the lock is set to false (OFF).

Note that even when AE is locked, the flash may be fired if the android.control.aeMode is ON_AUTO_FLASH / ON_ALWAYS_FLASH / ON_AUTO_FLASH_REDEYE.

When android.control.aeExposureCompensation is changed, even if the AE lock is ON, the camera device will still adjust its exposure value.

If AE precapture is triggered (see android.control.aePrecaptureTrigger) when AE is already locked, the camera device will not change the exposure time (android.sensor.exposureTime) and sensitivity (android.sensor.sensitivity) parameters. The flash may be fired if the android.control.aeMode is ON_AUTO_FLASH/ON_AUTO_FLASH_REDEYE and the scene is too dark. If the android.control.aeMode is ON_ALWAYS_FLASH, the scene may become overexposed. Similarly, AE precapture trigger CANCEL has no effect when AE is already locked.

When an AE precapture sequence is triggered, AE unlock will not be able to unlock the AE if AE is locked by the camera device internally during precapture metering sequence In other words, submitting requests with AE unlock has no effect for an ongoing precapture metering sequence. Otherwise, the precapture metering sequence will never succeed in a sequence of preview requests where AE lock is always set to false.

Since the camera device has a pipeline of in-flight requests, the settings that get locked do not necessarily correspond to the settings that were present in the latest capture result received from the camera device, since additional captures and AE updates may have occurred even before the result was sent out. If an application is switching between automatic and manual control and wishes to eliminate any flicker during the switch, the following procedure is recommended:

  1. Starting in auto-AE mode:
  2. Lock AE
  3. Wait for the first result to be output that has the AE locked
  4. Copy exposure settings from that result into a request, set the request to manual AE
  5. Submit the capture request, proceed to run manual AE as desired.

See android.control.aeState for AE lock related state transition details.

android.control.aeMode byte [public] [legacy]
  • OFF

    The camera device's autoexposure routine is disabled.

    The application-selected android.sensor.exposureTime, android.sensor.sensitivity and android.sensor.frameDuration are used by the camera device, along with android.flash.* fields, if there's a flash unit for this camera device.

    Note that auto-white balance (AWB) and auto-focus (AF) behavior is device dependent when AE is in OFF mode. To have consistent behavior across different devices, it is recommended to either set AWB and AF to OFF mode or lock AWB and AF before setting AE to OFF. See android.control.awbMode, android.control.afMode, android.control.awbLock, and android.control.afTrigger for more details.

    LEGACY devices do not support the OFF mode and will override attempts to use this value to ON.

  • ON

    The camera device's autoexposure routine is active, with no flash control.

    The application's values for android.sensor.exposureTime, android.sensor.sensitivity, and android.sensor.frameDuration are ignored. The application has control over the various android.flash.* fields.

  • ON_AUTO_FLASH

    Like ON, except that the camera device also controls the camera's flash unit, firing it in low-light conditions.

    The flash may be fired during a precapture sequence (triggered by android.control.aePrecaptureTrigger) and may be fired for captures for which the android.control.captureIntent field is set to STILL_CAPTURE

  • ON_ALWAYS_FLASH

    Like ON, except that the camera device also controls the camera's flash unit, always firing it for still captures.

    The flash may be fired during a precapture sequence (triggered by android.control.aePrecaptureTrigger) and will always be fired for captures for which the android.control.captureIntent field is set to STILL_CAPTURE

  • ON_AUTO_FLASH_REDEYE

    Like ON_AUTO_FLASH, but with automatic red eye reduction.

    If deemed necessary by the camera device, a red eye reduction flash will fire during the precapture sequence.

The desired mode for the camera device's auto-exposure routine.

android.control.aeAvailableModes

Details

This control is only effective if android.control.mode is AUTO.

When set to any of the ON modes, the camera device's auto-exposure routine is enabled, overriding the application's selected exposure time, sensor sensitivity, and frame duration (android.sensor.exposureTime, android.sensor.sensitivity, and android.sensor.frameDuration). If one of the FLASH modes is selected, the camera device's flash unit controls are also overridden.

The FLASH modes are only available if the camera device has a flash unit (android.flash.info.available is true).

If flash TORCH mode is desired, this field must be set to ON or OFF, and android.flash.mode set to TORCH.

When set to any of the ON modes, the values chosen by the camera device auto-exposure routine for the overridden fields for a given capture will be available in its CaptureResult.

android.control.aeRegions int32 x 5 x area_count [public as meteringRectangle]

List of metering areas to use for auto-exposure adjustment.

Pixel coordinates within android.sensor.info.activeArraySize

Coordinates must be between [(0,0), (width, height)) of android.sensor.info.activeArraySize

Details

Not available if android.control.maxRegionsAe is 0. Otherwise will always be present.

The maximum number of regions supported by the device is determined by the value of android.control.maxRegionsAe.

The coordinate system is based on the active pixel array, with (0,0) being the top-left pixel in the active pixel array, and (android.sensor.info.activeArraySize.width - 1, android.sensor.info.activeArraySize.height - 1) being the bottom-right pixel in the active pixel array.

The weight must be within [0, 1000], and represents a weight for every pixel in the area. This means that a large metering area with the same weight as a smaller area will have more effect in the metering result. Metering areas can partially overlap and the camera device will add the weights in the overlap region.

The weights are relative to weights of other exposure metering regions, so if only one region is used, all non-zero weights will have the same effect. A region with 0 weight is ignored.

If all regions have 0 weight, then no specific metering area needs to be used by the camera device.

If the metering region is outside the used android.scaler.cropRegion returned in capture result metadata, the camera device will ignore the sections outside the crop region and output only the intersection rectangle as the metering region in the result metadata. If the region is entirely outside the crop region, it will be ignored and not reported in the result metadata.

HAL Implementation Details

The HAL level representation of MeteringRectangle[] is a int[5 * area_count]. Every five elements represent a metering region of (xmin, ymin, xmax, ymax, weight). The rectangle is defined to be inclusive on xmin and ymin, but exclusive on xmax and ymax.

android.control.aeTargetFpsRange int32 x 2 [public as rangeInt] [legacy]

Range over which the auto-exposure routine can adjust the capture frame rate to maintain good exposure.

Frames per second (FPS)

Any of the entries in android.control.aeAvailableTargetFpsRanges

Details

Only constrains auto-exposure (AE) algorithm, not manual control of android.sensor.exposureTime and android.sensor.frameDuration.

android.control.aePrecaptureTrigger byte [public] [limited]
  • IDLE

    The trigger is idle.

  • START

    The precapture metering sequence will be started by the camera device.

    The exact effect of the precapture trigger depends on the current AE mode and state.

  • CANCEL

    The camera device will cancel any currently active or completed precapture metering sequence, the auto-exposure routine will return to its initial state.

Whether the camera device will trigger a precapture metering sequence when it processes this request.

Details

This entry is normally set to IDLE, or is not included at all in the request settings. When included and set to START, the camera device will trigger the auto-exposure (AE) precapture metering sequence.

When set to CANCEL, the camera device will cancel any active precapture metering trigger, and return to its initial AE state. If a precapture metering sequence is already completed, and the camera device has implicitly locked the AE for subsequent still capture, the CANCEL trigger will unlock the AE and return to its initial AE state.

The precapture sequence should be triggered before starting a high-quality still capture for final metering decisions to be made, and for firing pre-capture flash pulses to estimate scene brightness and required final capture flash power, when the flash is enabled.

Normally, this entry should be set to START for only a single request, and the application should wait until the sequence completes before starting a new one.

When a precapture metering sequence is finished, the camera device may lock the auto-exposure routine internally to be able to accurately expose the subsequent still capture image (android.control.captureIntent == STILL_CAPTURE). For this case, the AE may not resume normal scan if no subsequent still capture is submitted. To ensure that the AE routine restarts normal scan, the application should submit a request with android.control.aeLock == true, followed by a request with android.control.aeLock == false, if the application decides not to submit a still capture request after the precapture sequence completes. Alternatively, for API level 23 or newer devices, the CANCEL can be used to unlock the camera device internally locked AE if the application doesn't submit a still capture request after the AE precapture trigger. Note that, the CANCEL was added in API level 23, and must not be used in devices that have earlier API levels.

The exact effect of auto-exposure (AE) precapture trigger depends on the current AE mode and state; see android.control.aeState for AE precapture state transition details.

On LEGACY-level devices, the precapture trigger is not supported; capturing a high-resolution JPEG image will automatically trigger a precapture sequence before the high-resolution capture, including potentially firing a pre-capture flash.

Using the precapture trigger and the auto-focus trigger android.control.afTrigger simultaneously is allowed. However, since these triggers often require cooperation between the auto-focus and auto-exposure routines (for example, the may need to be enabled for a focus sweep), the camera device may delay acting on a later trigger until the previous trigger has been fully handled. This may lead to longer intervals between the trigger and changes to android.control.aeState indicating the start of the precapture sequence, for example.

If both the precapture and the auto-focus trigger are activated on the same request, then the camera device will complete them in the optimal order for that device.

HAL Implementation Details

The HAL must support triggering the AE precapture trigger while an AF trigger is active (and vice versa), or at the same time as the AF trigger. It is acceptable for the HAL to treat these as two consecutive triggers, for example handling the AF trigger and then the AE trigger. Or the HAL may choose to optimize the case with both triggers fired at once, to minimize the latency for converging both focus and exposure/flash usage.

android.control.aeState byte [public] [limited]
  • INACTIVE

    AE is off or recently reset.

    When a camera device is opened, it starts in this state. This is a transient state, the camera device may skip reporting this state in capture result.

  • SEARCHING

    AE doesn't yet have a good set of control values for the current scene.

    This is a transient state, the camera device may skip reporting this state in capture result.

  • CONVERGED

    AE has a good set of control values for the current scene.

  • LOCKED

    AE has been locked.

  • FLASH_REQUIRED

    AE has a good set of control values, but flash needs to be fired for good quality still capture.

  • PRECAPTURE

    AE has been asked to do a precapture sequence and is currently executing it.

    Precapture can be triggered through setting android.control.aePrecaptureTrigger to START. Currently active and completed (if it causes camera device internal AE lock) precapture metering sequence can be canceled through setting android.control.aePrecaptureTrigger to CANCEL.

    Once PRECAPTURE completes, AE will transition to CONVERGED or FLASH_REQUIRED as appropriate. This is a transient state, the camera device may skip reporting this state in capture result.

Current state of the auto-exposure (AE) algorithm.

Details

Switching between or enabling AE modes (android.control.aeMode) always resets the AE state to INACTIVE. Similarly, switching between android.control.mode, or android.control.sceneMode if android.control.mode == USE_SCENE_MODE resets all the algorithm states to INACTIVE.

The camera device can do several state transitions between two results, if it is allowed by the state transition table. For example: INACTIVE may never actually be seen in a result.

The state in the result is the state for this image (in sync with this image): if AE state becomes CONVERGED, then the image data associated with this result should be good to use.

Below are state transition tables for different AE modes.

State Transition Cause New State Notes
INACTIVE INACTIVE Camera device auto exposure algorithm is disabled

When android.control.aeMode is AE_MODE_ON_*:

State Transition Cause New State Notes
INACTIVE Camera device initiates AE scan SEARCHING Values changing
INACTIVE android.control.aeLock is ON LOCKED Values locked
SEARCHING Camera device finishes AE scan CONVERGED Good values, not changing
SEARCHING Camera device finishes AE scan FLASH_REQUIRED Converged but too dark w/o flash
SEARCHING android.control.aeLock is ON LOCKED Values locked
CONVERGED Camera device initiates AE scan SEARCHING Values changing
CONVERGED android.control.aeLock is ON LOCKED Values locked
FLASH_REQUIRED Camera device initiates AE scan SEARCHING Values changing
FLASH_REQUIRED android.control.aeLock is ON LOCKED Values locked
LOCKED android.control.aeLock is OFF SEARCHING Values not good after unlock
LOCKED android.control.aeLock is OFF CONVERGED Values good after unlock
LOCKED android.control.aeLock is OFF FLASH_REQUIRED Exposure good, but too dark
PRECAPTURE Sequence done. android.control.aeLock is OFF CONVERGED Ready for high-quality capture
PRECAPTURE Sequence done. android.control.aeLock is ON LOCKED Ready for high-quality capture
LOCKED aeLock is ON and aePrecaptureTrigger is START LOCKED Precapture trigger is ignored when AE is already locked
LOCKED aeLock is ON and aePrecaptureTrigger is CANCEL LOCKED Precapture trigger is ignored when AE is already locked
Any state (excluding LOCKED) android.control.aePrecaptureTrigger is START PRECAPTURE Start AE precapture metering sequence
Any state (excluding LOCKED) android.control.aePrecaptureTrigger is CANCEL INACTIVE Currently active precapture metering sequence is canceled

For the above table, the camera device may skip reporting any state changes that happen without application intervention (i.e. mode switch, trigger, locking). Any state that can be skipped in that manner is called a transient state.

For example, for above AE modes (AE_MODE_ON_*), in addition to the state transitions listed in above table, it is also legal for the camera device to skip one or more transient states between two results. See below table for examples:

State Transition Cause New State Notes
INACTIVE Camera device finished AE scan CONVERGED Values are already good, transient states are skipped by camera device.
Any state (excluding LOCKED) android.control.aePrecaptureTrigger is START, sequence done FLASH_REQUIRED Converged but too dark w/o flash after a precapture sequence, transient states are skipped by camera device.
Any state (excluding LOCKED) android.control.aePrecaptureTrigger is START, sequence done CONVERGED Converged after a precapture sequence, transient states are skipped by camera device.
Any state (excluding LOCKED) android.control.aePrecaptureTrigger is CANCEL, converged FLASH_REQUIRED Converged but too dark w/o flash after a precapture sequence is canceled, transient states are skipped by camera device.
Any state (excluding LOCKED) android.control.aePrecaptureTrigger is CANCEL, converged CONVERGED Converged after a precapture sequenceis canceled, transient states are skipped by camera device.
CONVERGED Camera device finished AE scan FLASH_REQUIRED Converged but too dark w/o flash after a new scan, transient states are skipped by camera device.
FLASH_REQUIRED Camera device finished AE scan CONVERGED Converged after a new scan, transient states are skipped by camera device.
android.control.afMode byte [public] [legacy]
  • OFF

    The auto-focus routine does not control the lens; android.lens.focusDistance is controlled by the application.

  • AUTO

    Basic automatic focus mode.

    In this mode, the lens does not move unless the autofocus trigger action is called. When that trigger is activated, AF will transition to ACTIVE_SCAN, then to the outcome of the scan (FOCUSED or NOT_FOCUSED).

    Always supported if lens is not fixed focus.

    Use android.lens.info.minimumFocusDistance to determine if lens is fixed-focus.

    Triggering AF_CANCEL resets the lens position to default, and sets the AF state to INACTIVE.

  • MACRO

    Close-up focusing mode.

    In this mode, the lens does not move unless the autofocus trigger action is called. When that trigger is activated, AF will transition to ACTIVE_SCAN, then to the outcome of the scan (FOCUSED or NOT_FOCUSED). This mode is optimized for focusing on objects very close to the camera.

    When that trigger is activated, AF will transition to ACTIVE_SCAN, then to the outcome of the scan (FOCUSED or NOT_FOCUSED). Triggering cancel AF resets the lens position to default, and sets the AF state to INACTIVE.

  • CONTINUOUS_VIDEO

    In this mode, the AF algorithm modifies the lens position continually to attempt to provide a constantly-in-focus image stream.

    The focusing behavior should be suitable for good quality video recording; typically this means slower focus movement and no overshoots. When the AF trigger is not involved, the AF algorithm should start in INACTIVE state, and then transition into PASSIVE_SCAN and PASSIVE_FOCUSED states as appropriate. When the AF trigger is activated, the algorithm should immediately transition into AF_FOCUSED or AF_NOT_FOCUSED as appropriate, and lock the lens position until a cancel AF trigger is received.

    Once cancel is received, the algorithm should transition back to INACTIVE and resume passive scan. Note that this behavior is not identical to CONTINUOUS_PICTURE, since an ongoing PASSIVE_SCAN must immediately be canceled.

  • CONTINUOUS_PICTURE

    In this mode, the AF algorithm modifies the lens position continually to attempt to provide a constantly-in-focus image stream.

    The focusing behavior should be suitable for still image capture; typically this means focusing as fast as possible. When the AF trigger is not involved, the AF algorithm should start in INACTIVE state, and then transition into PASSIVE_SCAN and PASSIVE_FOCUSED states as appropriate as it attempts to maintain focus. When the AF trigger is activated, the algorithm should finish its PASSIVE_SCAN if active, and then transition into AF_FOCUSED or AF_NOT_FOCUSED as appropriate, and lock the lens position until a cancel AF trigger is received.

    When the AF cancel trigger is activated, the algorithm should transition back to INACTIVE and then act as if it has just been started.

  • EDOF

    Extended depth of field (digital focus) mode.

    The camera device will produce images with an extended depth of field automatically; no special focusing operations need to be done before taking a picture.

    AF triggers are ignored, and the AF state will always be INACTIVE.

Whether auto-focus (AF) is currently enabled, and what mode it is set to.

android.control.afAvailableModes

Details

Only effective if android.control.mode = AUTO and the lens is not fixed focus (i.e. android.lens.info.minimumFocusDistance > 0). Also note that when android.control.aeMode is OFF, the behavior of AF is device dependent. It is recommended to lock AF by using android.control.afTrigger before setting android.control.aeMode to OFF, or set AF mode to OFF when AE is OFF.

If the lens is controlled by the camera device auto-focus algorithm, the camera device will report the current AF status in android.control.afState in result metadata.

HAL Implementation Details

When afMode is AUTO or MACRO, the lens must not move until an AF trigger is sent in a request (android.control.afTrigger == START). After an AF trigger, the afState will end up with either FOCUSED_LOCKED or NOT_FOCUSED_LOCKED state (see android.control.afState for detailed state transitions), which indicates that the lens is locked and will not move. If camera movement (e.g. tilting camera) causes the lens to move after the lens is locked, the HAL must compensate this movement appropriately such that the same focal plane remains in focus.

When afMode is one of the continuous auto focus modes, the HAL is free to start a AF scan whenever it's not locked. When the lens is locked after an AF trigger (see android.control.afState for detailed state transitions), the HAL should maintain the same lock behavior as above.

When afMode is OFF, the application controls focus manually. The accuracy of the focus distance control depends on the android.lens.info.focusDistanceCalibration. However, the lens must not move regardless of the camera movement for any focus distance manual control.

To put this in concrete terms, if the camera has lens elements which may move based on camera orientation or motion (e.g. due to gravity), then the HAL must drive the lens to remain in a fixed position invariant to the camera's orientation or motion, for example, by using accelerometer measurements in the lens control logic. This is a typical issue that will arise on camera modules with open-loop VCMs.

android.control.afRegions int32 x 5 x area_count [public as meteringRectangle]

List of metering areas to use for auto-focus.

Pixel coordinates within android.sensor.info.activeArraySize

Coordinates must be between [(0,0), (width, height)) of android.sensor.info.activeArraySize

Details

Not available if android.control.maxRegionsAf is 0. Otherwise will always be present.

The maximum number of focus areas supported by the device is determined by the value of android.control.maxRegionsAf.

The coordinate system is based on the active pixel array, with (0,0) being the top-left pixel in the active pixel array, and (android.sensor.info.activeArraySize.width - 1, android.sensor.info.activeArraySize.height - 1) being the bottom-right pixel in the active pixel array.

The weight must be within [0, 1000], and represents a weight for every pixel in the area. This means that a large metering area with the same weight as a smaller area will have more effect in the metering result. Metering areas can partially overlap and the camera device will add the weights in the overlap region.

The weights are relative to weights of other metering regions, so if only one region is used, all non-zero weights will have the same effect. A region with 0 weight is ignored.

If all regions have 0 weight, then no specific metering area needs to be used by the camera device.

If the metering region is outside the used android.scaler.cropRegion returned in capture result metadata, the camera device will ignore the sections outside the crop region and output only the intersection rectangle as the metering region in the result metadata. If the region is entirely outside the crop region, it will be ignored and not reported in the result metadata.

HAL Implementation Details

The HAL level representation of MeteringRectangle[] is a int[5 * area_count]. Every five elements represent a metering region of (xmin, ymin, xmax, ymax, weight). The rectangle is defined to be inclusive on xmin and ymin, but exclusive on xmax and ymax.

android.control.afTrigger byte [public] [legacy]
  • IDLE

    The trigger is idle.

  • START

    Autofocus will trigger now.

  • CANCEL

    Autofocus will return to its initial state, and cancel any currently active trigger.

Whether the camera device will trigger autofocus for this request.

Details

This entry is normally set to IDLE, or is not included at all in the request settings.

When included and set to START, the camera device will trigger the autofocus algorithm. If autofocus is disabled, this trigger has no effect.

When set to CANCEL, the camera device will cancel any active trigger, and return to its initial AF state.

Generally, applications should set this entry to START or CANCEL for only a single capture, and then return it to IDLE (or not set at all). Specifying START for multiple captures in a row means restarting the AF operation over and over again.

See android.control.afState for what the trigger means for each AF mode.

Using the autofocus trigger and the precapture trigger android.control.aePrecaptureTrigger simultaneously is allowed. However, since these triggers often require cooperation between the auto-focus and auto-exposure routines (for example, the may need to be enabled for a focus sweep), the camera device may delay acting on a later trigger until the previous trigger has been fully handled. This may lead to longer intervals between the trigger and changes to android.control.afState, for example.

HAL Implementation Details

The HAL must support triggering the AF trigger while an AE precapture trigger is active (and vice versa), or at the same time as the AE trigger. It is acceptable for the HAL to treat these as two consecutive triggers, for example handling the AF trigger and then the AE trigger. Or the HAL may choose to optimize the case with both triggers fired at once, to minimize the latency for converging both focus and exposure/flash usage.

android.control.afState byte [public] [legacy]
  • INACTIVE

    AF is off or has not yet tried to scan/been asked to scan.

    When a camera device is opened, it starts in this state. This is a transient state, the camera device may skip reporting this state in capture result.

  • PASSIVE_SCAN

    AF is currently performing an AF scan initiated the camera device in a continuous autofocus mode.

    Only used by CONTINUOUS_* AF modes. This is a transient state, the camera device may skip reporting this state in capture result.

  • PASSIVE_FOCUSED

    AF currently believes it is in focus, but may restart scanning at any time.

    Only used by CONTINUOUS_* AF modes. This is a transient state, the camera device may skip reporting this state in capture result.

  • ACTIVE_SCAN

    AF is performing an AF scan because it was triggered by AF trigger.

    Only used by AUTO or MACRO AF modes. This is a transient state, the camera device may skip reporting this state in capture result.

  • FOCUSED_LOCKED

    AF believes it is focused correctly and has locked focus.

    This state is reached only after an explicit START AF trigger has been sent (android.control.afTrigger), when good focus has been obtained.

    The lens will remain stationary until the AF mode (android.control.afMode) is changed or a new AF trigger is sent to the camera device (android.control.afTrigger).

  • NOT_FOCUSED_LOCKED

    AF has failed to focus successfully and has locked focus.

    This state is reached only after an explicit START AF trigger has been sent (android.control.afTrigger), when good focus cannot be obtained.

    The lens will remain stationary until the AF mode (android.control.afMode) is changed or a new AF trigger is sent to the camera device (android.control.afTrigger).

  • PASSIVE_UNFOCUSED

    AF finished a passive scan without finding focus, and may restart scanning at any time.

    Only used by CONTINUOUS_* AF modes. This is a transient state, the camera device may skip reporting this state in capture result.

    LEGACY camera devices do not support this state. When a passive scan has finished, it will always go to PASSIVE_FOCUSED.

Current state of auto-focus (AF) algorithm.

Details

Switching between or enabling AF modes (android.control.afMode) always resets the AF state to INACTIVE. Similarly, switching between android.control.mode, or android.control.sceneMode if android.control.mode == USE_SCENE_MODE resets all the algorithm states to INACTIVE.

The camera device can do several state transitions between two results, if it is allowed by the state transition table. For example: INACTIVE may never actually be seen in a result.

The state in the result is the state for this image (in sync with this image): if AF state becomes FOCUSED, then the image data associated with this result should be sharp.

Below are state transition tables for different AF modes.

When android.control.afMode is AF_MODE_OFF or AF_MODE_EDOF:

State Transition Cause New State Notes
INACTIVE INACTIVE Never changes

When android.control.afMode is AF_MODE_AUTO or AF_MODE_MACRO:

State Transition Cause New State Notes
INACTIVE AF_TRIGGER ACTIVE_SCAN Start AF sweep, Lens now moving
ACTIVE_SCAN AF sweep done FOCUSED_LOCKED Focused, Lens now locked
ACTIVE_SCAN AF sweep done NOT_FOCUSED_LOCKED Not focused, Lens now locked
ACTIVE_SCAN AF_CANCEL INACTIVE Cancel/reset AF, Lens now locked
FOCUSED_LOCKED AF_CANCEL INACTIVE Cancel/reset AF
FOCUSED_LOCKED AF_TRIGGER ACTIVE_SCAN Start new sweep, Lens now moving
NOT_FOCUSED_LOCKED AF_CANCEL INACTIVE Cancel/reset AF
NOT_FOCUSED_LOCKED AF_TRIGGER ACTIVE_SCAN Start new sweep, Lens now moving
Any state Mode change INACTIVE

For the above table, the camera device may skip reporting any state changes that happen without application intervention (i.e. mode switch, trigger, locking). Any state that can be skipped in that manner is called a transient state.

For example, for these AF modes (AF_MODE_AUTO and AF_MODE_MACRO), in addition to the state transitions listed in above table, it is also legal for the camera device to skip one or more transient states between two results. See below table for examples:

State Transition Cause New State Notes
INACTIVE AF_TRIGGER FOCUSED_LOCKED Focus is already good or good after a scan, lens is now locked.
INACTIVE AF_TRIGGER NOT_FOCUSED_LOCKED Focus failed after a scan, lens is now locked.
FOCUSED_LOCKED AF_TRIGGER FOCUSED_LOCKED Focus is already good or good after a scan, lens is now locked.
NOT_FOCUSED_LOCKED AF_TRIGGER FOCUSED_LOCKED Focus is good after a scan, lens is not locked.

When android.control.afMode is AF_MODE_CONTINUOUS_VIDEO:

State Transition Cause New State Notes
INACTIVE Camera device initiates new scan PASSIVE_SCAN Start AF scan, Lens now moving
INACTIVE AF_TRIGGER NOT_FOCUSED_LOCKED AF state query, Lens now locked
PASSIVE_SCAN Camera device completes current scan PASSIVE_FOCUSED End AF scan, Lens now locked
PASSIVE_SCAN Camera device fails current scan PASSIVE_UNFOCUSED End AF scan, Lens now locked
PASSIVE_SCAN AF_TRIGGER FOCUSED_LOCKED Immediate transition, if focus is good. Lens now locked
PASSIVE_SCAN AF_TRIGGER NOT_FOCUSED_LOCKED Immediate transition, if focus is bad. Lens now locked
PASSIVE_SCAN AF_CANCEL INACTIVE Reset lens position, Lens now locked
PASSIVE_FOCUSED Camera device initiates new scan PASSIVE_SCAN Start AF scan, Lens now moving
PASSIVE_UNFOCUSED Camera device initiates new scan PASSIVE_SCAN Start AF scan, Lens now moving
PASSIVE_FOCUSED AF_TRIGGER FOCUSED_LOCKED Immediate transition, lens now locked
PASSIVE_UNFOCUSED AF_TRIGGER NOT_FOCUSED_LOCKED Immediate transition, lens now locked
FOCUSED_LOCKED AF_TRIGGER FOCUSED_LOCKED No effect
FOCUSED_LOCKED AF_CANCEL INACTIVE Restart AF scan
NOT_FOCUSED_LOCKED AF_TRIGGER NOT_FOCUSED_LOCKED No effect
NOT_FOCUSED_LOCKED AF_CANCEL INACTIVE Restart AF scan

When android.control.afMode is AF_MODE_CONTINUOUS_PICTURE:

State Transition Cause New State Notes
INACTIVE Camera device initiates new scan PASSIVE_SCAN Start AF scan, Lens now moving
INACTIVE AF_TRIGGER NOT_FOCUSED_LOCKED AF state query, Lens now locked
PASSIVE_SCAN Camera device completes current scan PASSIVE_FOCUSED End AF scan, Lens now locked
PASSIVE_SCAN Camera device fails current scan PASSIVE_UNFOCUSED End AF scan, Lens now locked
PASSIVE_SCAN AF_TRIGGER FOCUSED_LOCKED Eventual transition once the focus is good. Lens now locked
PASSIVE_SCAN AF_TRIGGER NOT_FOCUSED_LOCKED Eventual transition if cannot find focus. Lens now locked
PASSIVE_SCAN AF_CANCEL INACTIVE Reset lens position, Lens now locked
PASSIVE_FOCUSED Camera device initiates new scan PASSIVE_SCAN Start AF scan, Lens now moving
PASSIVE_UNFOCUSED Camera device initiates new scan PASSIVE_SCAN Start AF scan, Lens now moving
PASSIVE_FOCUSED AF_TRIGGER FOCUSED_LOCKED Immediate trans. Lens now locked
PASSIVE_UNFOCUSED AF_TRIGGER NOT_FOCUSED_LOCKED Immediate trans. Lens now locked
FOCUSED_LOCKED AF_TRIGGER FOCUSED_LOCKED No effect
FOCUSED_LOCKED AF_CANCEL INACTIVE Restart AF scan
NOT_FOCUSED_LOCKED AF_TRIGGER NOT_FOCUSED_LOCKED No effect
NOT_FOCUSED_LOCKED AF_CANCEL INACTIVE Restart AF scan

When switch between AF_MODE_CONTINUOUS_* (CAF modes) and AF_MODE_AUTO/AF_MODE_MACRO (AUTO modes), the initial INACTIVE or PASSIVE_SCAN states may be skipped by the camera device. When a trigger is included in a mode switch request, the trigger will be evaluated in the context of the new mode in the request. See below table for examples:

State Transition Cause New State Notes
any state CAF-->AUTO mode switch INACTIVE Mode switch without trigger, initial state must be INACTIVE
any state CAF-->AUTO mode switch with AF_TRIGGER trigger-reachable states from INACTIVE Mode switch with trigger, INACTIVE is skipped
any state AUTO-->CAF mode switch passively reachable states from INACTIVE Mode switch without trigger, passive transient state is skipped
android.control.afTriggerId int32 [system] [deprecated]

The ID sent with the latest CAMERA2_TRIGGER_AUTOFOCUS call

Deprecated. Do not use.

Details

Must be 0 if no CAMERA2_TRIGGER_AUTOFOCUS trigger received yet by HAL. Always updated even if AF algorithm ignores the trigger

android.control.awbLock byte [public as boolean] [legacy]
  • OFF

    Auto-white balance lock is disabled; the AWB algorithm is free to update its parameters if in AUTO mode.

  • ON

    Auto-white balance lock is enabled; the AWB algorithm will not update its parameters while the lock is active.

Whether auto-white balance (AWB) is currently locked to its latest calculated values.

Details

When set to true (ON), the AWB algorithm is locked to its latest parameters, and will not change color balance settings until the lock is set to false (OFF).

Since the camera device has a pipeline of in-flight requests, the settings that get locked do not necessarily correspond to the settings that were present in the latest capture result received from the camera device, since additional captures and AWB updates may have occurred even before the result was sent out. If an application is switching between automatic and manual control and wishes to eliminate any flicker during the switch, the following procedure is recommended:

  1. Starting in auto-AWB mode:
  2. Lock AWB
  3. Wait for the first result to be output that has the AWB locked
  4. Copy AWB settings from that result into a request, set the request to manual AWB
  5. Submit the capture request, proceed to run manual AWB as desired.

Note that AWB lock is only meaningful when android.control.awbMode is in the AUTO mode; in other modes, AWB is already fixed to a specific setting.

Some LEGACY devices may not support ON; the value is then overridden to OFF.

android.control.awbMode byte [public] [legacy]
  • OFF

    The camera device's auto-white balance routine is disabled.

    The application-selected color transform matrix (android.colorCorrection.transform) and gains (android.colorCorrection.gains) are used by the camera device for manual white balance control.

  • AUTO

    The camera device's auto-white balance routine is active.

    The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request.

  • INCANDESCENT

    The camera device's auto-white balance routine is disabled; the camera device uses incandescent light as the assumed scene illumination for white balance.

    While the exact white balance transforms are up to the camera device, they will approximately match the CIE standard illuminant A.

    The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request.

  • FLUORESCENT

    The camera device's auto-white balance routine is disabled; the camera device uses fluorescent light as the assumed scene illumination for white balance.

    While the exact white balance transforms are up to the camera device, they will approximately match the CIE standard illuminant F2.

    The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request.

  • WARM_FLUORESCENT

    The camera device's auto-white balance routine is disabled; the camera device uses warm fluorescent light as the assumed scene illumination for white balance.

    While the exact white balance transforms are up to the camera device, they will approximately match the CIE standard illuminant F4.

    The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request.

  • DAYLIGHT

    The camera device's auto-white balance routine is disabled; the camera device uses daylight light as the assumed scene illumination for white balance.

    While the exact white balance transforms are up to the camera device, they will approximately match the CIE standard illuminant D65.

    The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request.

  • CLOUDY_DAYLIGHT

    The camera device's auto-white balance routine is disabled; the camera device uses cloudy daylight light as the assumed scene illumination for white balance.

    The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request.

  • TWILIGHT

    The camera device's auto-white balance routine is disabled; the camera device uses twilight light as the assumed scene illumination for white balance.

    The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request.

  • SHADE

    The camera device's auto-white balance routine is disabled; the camera device uses shade light as the assumed scene illumination for white balance.

    The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request.

Whether auto-white balance (AWB) is currently setting the color transform fields, and what its illumination target is.

android.control.awbAvailableModes

Details

This control is only effective if android.control.mode is AUTO.

When set to the ON mode, the camera device's auto-white balance routine is enabled, overriding the application's selected android.colorCorrection.transform, android.colorCorrection.gains and android.colorCorrection.mode. Note that when android.control.aeMode is OFF, the behavior of AWB is device dependent. It is recommened to also set AWB mode to OFF or lock AWB by using android.control.awbLock before setting AE mode to OFF.

When set to the OFF mode, the camera device's auto-white balance routine is disabled. The application manually controls the white balance by android.colorCorrection.transform, android.colorCorrection.gains and android.colorCorrection.mode.

When set to any other modes, the camera device's auto-white balance routine is disabled. The camera device uses each particular illumination target for white balance adjustment. The application's values for android.colorCorrection.transform, android.colorCorrection.gains and android.colorCorrection.mode are ignored.

android.control.awbRegions int32 x 5 x area_count [public as meteringRectangle]

List of metering areas to use for auto-white-balance illuminant estimation.

Pixel coordinates within android.sensor.info.activeArraySize

Coordinates must be between [(0,0), (width, height)) of android.sensor.info.activeArraySize

Details

Not available if android.control.maxRegionsAwb is 0. Otherwise will always be present.

The maximum number of regions supported by the device is determined by the value of android.control.maxRegionsAwb.

The coordinate system is based on the active pixel array, with (0,0) being the top-left pixel in the active pixel array, and (android.sensor.info.activeArraySize.width - 1, android.sensor.info.activeArraySize.height - 1) being the bottom-right pixel in the active pixel array.

The weight must range from 0 to 1000, and represents a weight for every pixel in the area. This means that a large metering area with the same weight as a smaller area will have more effect in the metering result. Metering areas can partially overlap and the camera device will add the weights in the overlap region.

The weights are relative to weights of other white balance metering regions, so if only one region is used, all non-zero weights will have the same effect. A region with 0 weight is ignored.

If all regions have 0 weight, then no specific metering area needs to be used by the camera device.

If the metering region is outside the used android.scaler.cropRegion returned in capture result metadata, the camera device will ignore the sections outside the crop region and output only the intersection rectangle as the metering region in the result metadata. If the region is entirely outside the crop region, it will be ignored and not reported in the result metadata.

HAL Implementation Details

The HAL level representation of MeteringRectangle[] is a int[5 * area_count]. Every five elements represent a metering region of (xmin, ymin, xmax, ymax, weight). The rectangle is defined to be inclusive on xmin and ymin, but exclusive on xmax and ymax.

android.control.captureIntent byte [public] [legacy]
  • CUSTOM

    The goal of this request doesn't fall into the other categories. The camera device will default to preview-like behavior.

  • PREVIEW

    This request is for a preview-like use case.

    The precapture trigger may be used to start off a metering w/flash sequence.

  • STILL_CAPTURE

    This request is for a still capture-type use case.

    If the flash unit is under automatic control, it may fire as needed.

  • VIDEO_RECORD

    This request is for a video recording use case.

  • VIDEO_SNAPSHOT

    This request is for a video snapshot (still image while recording video) use case.

    The camera device should take the highest-quality image possible (given the other settings) without disrupting the frame rate of video recording.

  • ZERO_SHUTTER_LAG

    This request is for a ZSL usecase; the application will stream full-resolution images and reprocess one or several later for a final capture.

  • MANUAL

    This request is for manual capture use case where the applications want to directly control the capture parameters.

    For example, the application may wish to manually control android.sensor.exposureTime, android.sensor.sensitivity, etc.

Information to the camera device 3A (auto-exposure, auto-focus, auto-white balance) routines about the purpose of this capture, to help the camera device to decide optimal 3A strategy.

Details

This control (except for MANUAL) is only effective if android.control.mode != OFF and any 3A routine is active.

ZERO_SHUTTER_LAG will be supported if android.request.availableCapabilities contains PRIVATE_REPROCESSING or YUV_REPROCESSING. MANUAL will be supported if android.request.availableCapabilities contains MANUAL_SENSOR. Other intent values are always supported.

android.control.awbState byte [public] [limited]
  • INACTIVE

    AWB is not in auto mode, or has not yet started metering.

    When a camera device is opened, it starts in this state. This is a transient state, the camera device may skip reporting this state in capture result.

  • SEARCHING

    AWB doesn't yet have a good set of control values for the current scene.

    This is a transient state, the camera device may skip reporting this state in capture result.

  • CONVERGED

    AWB has a good set of control values for the current scene.

  • LOCKED

    AWB has been locked.

Current state of auto-white balance (AWB) algorithm.

Details

Switching between or enabling AWB modes (android.control.awbMode) always resets the AWB state to INACTIVE. Similarly, switching between android.control.mode, or android.control.sceneMode if android.control.mode == USE_SCENE_MODE resets all the algorithm states to INACTIVE.

The camera device can do several state transitions between two results, if it is allowed by the state transition table. So INACTIVE may never actually be seen in a result.

The state in the result is the state for this image (in sync with this image): if AWB state becomes CONVERGED, then the image data associated with this result should be good to use.

Below are state transition tables for different AWB modes.

When android.control.awbMode != AWB_MODE_AUTO:

State Transition Cause New State Notes
INACTIVE INACTIVE Camera device auto white balance algorithm is disabled

When android.control.awbMode is AWB_MODE_AUTO:

State Transition Cause New State Notes
INACTIVE Camera device initiates AWB scan SEARCHING Values changing
INACTIVE android.control.awbLock is ON LOCKED Values locked
SEARCHING Camera device finishes AWB scan CONVERGED Good values, not changing
SEARCHING android.control.awbLock is ON LOCKED Values locked
CONVERGED Camera device initiates AWB scan SEARCHING Values changing
CONVERGED android.control.awbLock is ON LOCKED Values locked
LOCKED android.control.awbLock is OFF SEARCHING Values not good after unlock

For the above table, the camera device may skip reporting any state changes that happen without application intervention (i.e. mode switch, trigger, locking). Any state that can be skipped in that manner is called a transient state.

For example, for this AWB mode (AWB_MODE_AUTO), in addition to the state transitions listed in above table, it is also legal for the camera device to skip one or more transient states between two results. See below table for examples:

State Transition Cause New State Notes
INACTIVE Camera device finished AWB scan CONVERGED Values are already good, transient states are skipped by camera device.
LOCKED android.control.awbLock is OFF CONVERGED Values good after unlock, transient states are skipped by camera device.
android.control.effectMode byte [public] [legacy]
  • OFF

    No color effect will be applied.

  • MONO [optional]

    A "monocolor" effect where the image is mapped into a single color.

    This will typically be grayscale.

  • NEGATIVE [optional]

    A "photo-negative" effect where the image's colors are inverted.

  • SOLARIZE [optional]

    A "solarisation" effect (Sabattier effect) where the image is wholly or partially reversed in tone.

  • SEPIA [optional]

    A "sepia" effect where the image is mapped into warm gray, red, and brown tones.

  • POSTERIZE [optional]

    A "posterization" effect where the image uses discrete regions of tone rather than a continuous gradient of tones.

  • WHITEBOARD [optional]

    A "whiteboard" effect where the image is typically displayed as regions of white, with black or grey details.

  • BLACKBOARD [optional]

    A "blackboard" effect where the image is typically displayed as regions of black, with white or grey details.

  • AQUA [optional]

    An "aqua" effect where a blue hue is added to the image.

A special color effect to apply.

android.control.availableEffects

Details

When this mode is set, a color effect will be applied to images produced by the camera device. The interpretation and implementation of these color effects is left to the implementor of the camera device, and should not be depended on to be consistent (or present) across all devices.

android.control.mode byte [public] [legacy]
  • OFF

    Full application control of pipeline.

    All control by the device's metering and focusing (3A) routines is disabled, and no other settings in android.control.* have any effect, except that android.control.captureIntent may be used by the camera device to select post-processing values for processing blocks that do not allow for manual control, or are not exposed by the camera API.

    However, the camera device's 3A routines may continue to collect statistics and update their internal state so that when control is switched to AUTO mode, good control values can be immediately applied.

  • AUTO

    Use settings for each individual 3A routine.

    Manual control of capture parameters is disabled. All controls in android.control.* besides sceneMode take effect.

  • USE_SCENE_MODE [optional]

    Use a specific scene mode.

    Enabling this disables control.aeMode, control.awbMode and control.afMode controls; the camera device will ignore those settings while USE_SCENE_MODE is active (except for FACE_PRIORITY scene mode). Other control entries are still active. This setting can only be used if scene mode is supported (i.e. android.control.availableSceneModes contain some modes other than DISABLED).

  • OFF_KEEP_STATE [optional]

    Same as OFF mode, except that this capture will not be used by camera device background auto-exposure, auto-white balance and auto-focus algorithms (3A) to update their statistics.

    Specifically, the 3A routines are locked to the last values set from a request with AUTO, OFF, or USE_SCENE_MODE, and any statistics or state updates collected from manual captures with OFF_KEEP_STATE will be discarded by the camera device.

Overall mode of 3A (auto-exposure, auto-white-balance, auto-focus) control routines.

android.control.availableModes

Details

This is a top-level 3A control switch. When set to OFF, all 3A control by the camera device is disabled. The application must set the fields for capture parameters itself.

When set to AUTO, the individual algorithm controls in android.control.* are in effect, such as android.control.afMode.

When set to USE_SCENE_MODE, the individual controls in android.control.* are mostly disabled, and the camera device implements one of the scene mode settings (such as ACTION, SUNSET, or PARTY) as it wishes. The camera device scene mode 3A settings are provided by capture results.

When set to OFF_KEEP_STATE, it is similar to OFF mode, the only difference is that this frame will not be used by camera device background 3A statistics update, as if this frame is never captured. This mode can be used in the scenario where the application doesn't want a 3A manual control capture to affect the subsequent auto 3A capture results.

android.control.sceneMode byte [public] [legacy]
  • DISABLED 0

    Indicates that no scene modes are set for a given capture request.

  • FACE_PRIORITY

    If face detection support exists, use face detection data for auto-focus, auto-white balance, and auto-exposure routines.

    If face detection statistics are disabled (i.e. android.statistics.faceDetectMode is set to OFF), this should still operate correctly (but will not return face detection statistics to the framework).

    Unlike the other scene modes, android.control.aeMode, android.control.awbMode, and android.control.afMode remain active when FACE_PRIORITY is set.

  • ACTION [optional]

    Optimized for photos of quickly moving objects.

    Similar to SPORTS.

  • PORTRAIT [optional]

    Optimized for still photos of people.

  • LANDSCAPE [optional]

    Optimized for photos of distant macroscopic objects.

  • NIGHT [optional]

    Optimized for low-light settings.

  • NIGHT_PORTRAIT [optional]

    Optimized for still photos of people in low-light settings.

  • THEATRE [optional]

    Optimized for dim, indoor settings where flash must remain off.

  • BEACH [optional]

    Optimized for bright, outdoor beach settings.

  • SNOW [optional]

    Optimized for bright, outdoor settings containing snow.

  • SUNSET [optional]

    Optimized for scenes of the setting sun.

  • STEADYPHOTO [optional]

    Optimized to avoid blurry photos due to small amounts of device motion (for example: due to hand shake).

  • FIREWORKS [optional]

    Optimized for nighttime photos of fireworks.

  • SPORTS [optional]

    Optimized for photos of quickly moving people.

    Similar to ACTION.

  • PARTY [optional]

    Optimized for dim, indoor settings with multiple moving people.

  • CANDLELIGHT [optional]

    Optimized for dim settings where the main light source is a flame.

  • BARCODE [optional]

    Optimized for accurately capturing a photo of barcode for use by camera applications that wish to read the barcode value.

  • HIGH_SPEED_VIDEO [deprecated] [optional]

    This is deprecated, please use CameraDevice#createConstrainedHighSpeedCaptureSession and CameraConstrainedHighSpeedCaptureSession#createHighSpeedRequestList for high speed video recording.

    Optimized for high speed video recording (frame rate >=60fps) use case.

    The supported high speed video sizes and fps ranges are specified in android.control.availableHighSpeedVideoConfigurations. To get desired output frame rates, the application is only allowed to select video size and fps range combinations listed in this static metadata. The fps range can be control via android.control.aeTargetFpsRange.

    In this mode, the camera device will override aeMode, awbMode, and afMode to ON, ON, and CONTINUOUS_VIDEO, respectively. All post-processing block mode controls will be overridden to be FAST. Therefore, no manual control of capture and post-processing parameters is possible. All other controls operate the same as when android.control.mode == AUTO. This means that all other android.control.* fields continue to work, such as

    Outside of android.control.*, the following controls will work:

    For high speed recording use case, the actual maximum supported frame rate may be lower than what camera can output, depending on the destination Surfaces for the image data. For example, if the destination surface is from video encoder, the application need check if the video encoder is capable of supporting the high frame rate for a given video size, or it will end up with lower recording frame rate. If the destination surface is from preview window, the preview frame rate will be bounded by the screen refresh rate.

    The camera device will only support up to 2 output high speed streams (processed non-stalling format defined in android.request.maxNumOutputStreams) in this mode. This control will be effective only if all of below conditions are true:

    When above conditions are NOT satistied, the controls of this mode and android.control.aeTargetFpsRange will be ignored by the camera device, the camera device will fall back to android.control.mode == AUTO, and the returned capture result metadata will give the fps range choosen by the camera device.

    Switching into or out of this mode may trigger some camera ISP/sensor reconfigurations, which may introduce extra latency. It is recommended that the application avoids unnecessary scene mode switch as much as possible.

  • HDR [optional]

    Turn on a device-specific high dynamic range (HDR) mode.

    In this scene mode, the camera device captures images that keep a larger range of scene illumination levels visible in the final image. For example, when taking a picture of a object in front of a bright window, both the object and the scene through the window may be visible when using HDR mode, while in normal AUTO mode, one or the other may be poorly exposed. As a tradeoff, HDR mode generally takes much longer to capture a single image, has no user control, and may have other artifacts depending on the HDR method used.

    Therefore, HDR captures operate at a much slower rate than regular captures.

    In this mode, on LIMITED or FULL devices, when a request is made with a android.control.captureIntent of STILL_CAPTURE, the camera device will capture an image using a high dynamic range capture technique. On LEGACY devices, captures that target a JPEG-format output will be captured with HDR, and the capture intent is not relevant.

    The HDR capture may involve the device capturing a burst of images internally and combining them into one, or it may involve the device using specialized high dynamic range capture hardware. In all cases, a single image is produced in response to a capture request submitted while in HDR mode.

    Since substantial post-processing is generally needed to produce an HDR image, only YUV and JPEG outputs are supported for LIMITED/FULL device HDR captures, and only JPEG outputs are supported for LEGACY HDR captures. Using a RAW output for HDR capture is not supported.

  • FACE_PRIORITY_LOW_LIGHT [optional] [hidden]

    Same as FACE_PRIORITY scene mode, except that the camera device will choose higher sensitivity values (android.sensor.sensitivity) under low light conditions.

    The camera device may be tuned to expose the images in a reduced sensitivity range to produce the best quality images. For example, if the android.sensor.info.sensitivityRange gives range of [100, 1600], the camera device auto-exposure routine tuning process may limit the actual exposure sensitivity range to [100, 1200] to ensure that the noise level isn't exessive in order to preserve the image quality. Under this situation, the image under low light may be under-exposed when the sensor max exposure time (bounded by the android.control.aeTargetFpsRange when android.control.aeMode is one of the ON_* modes) and effective max sensitivity are reached. This scene mode allows the camera device auto-exposure routine to increase the sensitivity up to the max sensitivity specified by android.sensor.info.sensitivityRange when the scene is too dark and the max exposure time is reached. The captured images may be noisier compared with the images captured in normal FACE_PRIORITY mode; therefore, it is recommended that the application only use this scene mode when it is capable of reducing the noise level of the captured images.

    Unlike the other scene modes, android.control.aeMode, android.control.awbMode, and android.control.afMode remain active when FACE_PRIORITY_LOW_LIGHT is set.

Control for which scene mode is currently active.

android.control.availableSceneModes

Details

Scene modes are custom camera modes optimized for a certain set of conditions and capture settings.

This is the mode that that is active when android.control.mode == USE_SCENE_MODE. Aside from FACE_PRIORITY, these modes will disable android.control.aeMode, android.control.awbMode, and android.control.afMode while in use.

The interpretation and implementation of these scene modes is left to the implementor of the camera device. Their behavior will not be consistent across all devices, and any given device may only implement a subset of these modes.

HAL Implementation Details

HAL implementations that include scene modes are expected to provide the per-scene settings to use for android.control.aeMode, android.control.awbMode, and android.control.afMode in android.control.sceneModeOverrides.

For HIGH_SPEED_VIDEO mode, if it is included in android.control.availableSceneModes, the HAL must list supported video size and fps range in android.control.availableHighSpeedVideoConfigurations. For a given size, e.g. 1280x720, if the HAL has two different sensor configurations for normal streaming mode and high speed streaming, when this scene mode is set/reset in a sequence of capture requests, the HAL may have to switch between different sensor modes. This mode is deprecated in HAL3.3, to support high speed video recording, please implement android.control.availableHighSpeedVideoConfigurations and CONSTRAINED_HIGH_SPEED_VIDEO capbility defined in android.request.availableCapabilities.

android.control.videoStabilizationMode byte [public] [legacy]
  • OFF

    Video stabilization is disabled.

  • ON

    Video stabilization is enabled.

Whether video stabilization is active.

Details

Video stabilization automatically warps images from the camera in order to stabilize motion between consecutive frames.

If enabled, video stabilization can modify the android.scaler.cropRegion to keep the video stream stabilized.

Switching between different video stabilization modes may take several frames to initialize, the camera device will report the current mode in capture result metadata. For example, When "ON" mode is requested, the video stabilization modes in the first several capture results may still be "OFF", and it will become "ON" when the initialization is done.

In addition, not all recording sizes or frame rates may be supported for stabilization by a device that reports stabilization support. It is guaranteed that an output targeting a MediaRecorder or MediaCodec will be stabilized if the recording resolution is less than or equal to 1920 x 1080 (width less than or equal to 1920, height less than or equal to 1080), and the recording frame rate is less than or equal to 30fps. At other sizes, the CaptureResult android.control.videoStabilizationMode field will return OFF if the recording output is not stabilized, or if there are no output Surface types that can be stabilized.

If a camera device supports both this mode and OIS (android.lens.opticalStabilizationMode), turning both modes on may produce undesirable interaction, so it is recommended not to enable both at the same time.

demosaic
controls
Property Name Type Description Units Range Tags
android.demosaic.mode byte [system]
  • FAST

    Minimal or no slowdown of frame rate compared to Bayer RAW output.

  • HIGH_QUALITY

    Improved processing quality but the frame rate might be slowed down relative to raw output.

Controls the quality of the demosaicing processing.

edge
controls
Property Name Type Description Units Range Tags
android.edge.mode byte [public] [full]
  • OFF

    No edge enhancement is applied.

  • FAST

    Apply edge enhancement at a quality level that does not slow down frame rate relative to sensor output. It may be the same as OFF if edge enhancement will slow down frame rate relative to sensor.

  • HIGH_QUALITY

    Apply high-quality edge enhancement, at a cost of possibly reduced output frame rate.

  • ZERO_SHUTTER_LAG [optional]

    Edge enhancement is applied at different levels for different output streams, based on resolution. Streams at maximum recording resolution (see CameraDevice#createCaptureSession) or below have edge enhancement applied, while higher-resolution streams have no edge enhancement applied. The level of edge enhancement for low-resolution streams is tuned so that frame rate is not impacted, and the quality is equal to or better than FAST (since it is only applied to lower-resolution outputs, quality may improve from FAST).

    This mode is intended to be used by applications operating in a zero-shutter-lag mode with YUV or PRIVATE reprocessing, where the application continuously captures high-resolution intermediate buffers into a circular buffer, from which a final image is produced via reprocessing when a user takes a picture. For such a use case, the high-resolution buffers must not have edge enhancement applied to maximize efficiency of preview and to avoid double-applying enhancement when reprocessed, while low-resolution buffers (used for recording or preview, generally) need edge enhancement applied for reasonable preview quality.

    This mode is guaranteed to be supported by devices that support either the YUV_REPROCESSING or PRIVATE_REPROCESSING capabilities (android.request.availableCapabilities lists either of those capabilities) and it will be the default mode for CAMERA3_TEMPLATE_ZERO_SHUTTER_LAG template.

Operation mode for edge enhancement.

android.edge.availableEdgeModes

Details

Edge enhancement improves sharpness and details in the captured image. OFF means no enhancement will be applied by the camera device.

FAST/HIGH_QUALITY both mean camera device determined enhancement will be applied. HIGH_QUALITY mode indicates that the camera device will use the highest-quality enhancement algorithms, even if it slows down capture rate. FAST means the camera device will not slow down capture rate when applying edge enhancement. FAST may be the same as OFF if edge enhancement will slow down capture rate. Every output stream will have a similar amount of enhancement applied.

ZERO_SHUTTER_LAG is meant to be used by applications that maintain a continuous circular buffer of high-resolution images during preview and reprocess image(s) from that buffer into a final capture when triggered by the user. In this mode, the camera device applies edge enhancement to low-resolution streams (below maximum recording resolution) to maximize preview quality, but does not apply edge enhancement to high-resolution streams, since those will be reprocessed later if necessary.

For YUV_REPROCESSING, these FAST/HIGH_QUALITY modes both mean that the camera device will apply FAST/HIGH_QUALITY YUV-domain edge enhancement, respectively. The camera device may adjust its internal edge enhancement parameters for best image quality based on the android.reprocess.effectiveExposureFactor, if it is set.

HAL Implementation Details

For YUV_REPROCESSING The HAL can use android.reprocess.effectiveExposureFactor to adjust the internal edge enhancement reduction parameters appropriately to get the best quality images.

android.edge.strength byte [system]

Control the amount of edge enhancement applied to the images

1-10; 10 is maximum sharpening
static
Property Name Type Description Units Range Tags
android.edge.availableEdgeModes byte x n [public as enumList] [full]
list of enums

List of edge enhancement modes for android.edge.mode that are supported by this camera device.

Any value listed in android.edge.mode

Details

Full-capability camera devices must always support OFF; camera devices that support YUV_REPROCESSING or PRIVATE_REPROCESSING will list ZERO_SHUTTER_LAG; all devices will list FAST.

HAL Implementation Details

HAL must support both FAST and HIGH_QUALITY if edge enhancement control is available on the camera device, but the underlying implementation can be the same for both modes. That is, if the highest quality implementation on the camera device does not slow down capture rate, then FAST and HIGH_QUALITY will generate the same output.

dynamic
Property Name Type Description Units Range Tags
android.edge.mode byte [public] [full]
  • OFF

    No edge enhancement is applied.

  • FAST

    Apply edge enhancement at a quality level that does not slow down frame rate relative to sensor output. It may be the same as OFF if edge enhancement will slow down frame rate relative to sensor.

  • HIGH_QUALITY

    Apply high-quality edge enhancement, at a cost of possibly reduced output frame rate.

  • ZERO_SHUTTER_LAG [optional]

    Edge enhancement is applied at different levels for different output streams, based on resolution. Streams at maximum recording resolution (see CameraDevice#createCaptureSession) or below have edge enhancement applied, while higher-resolution streams have no edge enhancement applied. The level of edge enhancement for low-resolution streams is tuned so that frame rate is not impacted, and the quality is equal to or better than FAST (since it is only applied to lower-resolution outputs, quality may improve from FAST).

    This mode is intended to be used by applications operating in a zero-shutter-lag mode with YUV or PRIVATE reprocessing, where the application continuously captures high-resolution intermediate buffers into a circular buffer, from which a final image is produced via reprocessing when a user takes a picture. For such a use case, the high-resolution buffers must not have edge enhancement applied to maximize efficiency of preview and to avoid double-applying enhancement when reprocessed, while low-resolution buffers (used for recording or preview, generally) need edge enhancement applied for reasonable preview quality.

    This mode is guaranteed to be supported by devices that support either the YUV_REPROCESSING or PRIVATE_REPROCESSING capabilities (android.request.availableCapabilities lists either of those capabilities) and it will be the default mode for CAMERA3_TEMPLATE_ZERO_SHUTTER_LAG template.

Operation mode for edge enhancement.

android.edge.availableEdgeModes

Details

Edge enhancement improves sharpness and details in the captured image. OFF means no enhancement will be applied by the camera device.

FAST/HIGH_QUALITY both mean camera device determined enhancement will be applied. HIGH_QUALITY mode indicates that the camera device will use the highest-quality enhancement algorithms, even if it slows down capture rate. FAST means the camera device will not slow down capture rate when applying edge enhancement. FAST may be the same as OFF if edge enhancement will slow down capture rate. Every output stream will have a similar amount of enhancement applied.

ZERO_SHUTTER_LAG is meant to be used by applications that maintain a continuous circular buffer of high-resolution images during preview and reprocess image(s) from that buffer into a final capture when triggered by the user. In this mode, the camera device applies edge enhancement to low-resolution streams (below maximum recording resolution) to maximize preview quality, but does not apply edge enhancement to high-resolution streams, since those will be reprocessed later if necessary.

For YUV_REPROCESSING, these FAST/HIGH_QUALITY modes both mean that the camera device will apply FAST/HIGH_QUALITY YUV-domain edge enhancement, respectively. The camera device may adjust its internal edge enhancement parameters for best image quality based on the android.reprocess.effectiveExposureFactor, if it is set.

HAL Implementation Details

For YUV_REPROCESSING The HAL can use android.reprocess.effectiveExposureFactor to adjust the internal edge enhancement reduction parameters appropriately to get the best quality images.

flash
controls
Property Name Type Description Units Range Tags
android.flash.firingPower byte [system]

Power for flash firing/torch

10 is max power; 0 is no flash. Linear

0 - 10

Details

Power for snapshot may use a different scale than for torch mode. Only one entry for torch mode will be used

android.flash.firingTime int64 [system]

Firing time of flash relative to start of exposure

nanoseconds

0-(exposure time-flash duration)

Details

Clamped to (0, exposure time - flash duration).

android.flash.mode byte [public] [legacy]
  • OFF

    Do not fire the flash for this capture.

  • SINGLE

    If the flash is available and charged, fire flash for this capture.

  • TORCH

    Transition flash to continuously on.

The desired mode for for the camera device's flash control.

Details

This control is only effective when flash unit is available (android.flash.info.available == true).

When this control is used, the android.control.aeMode must be set to ON or OFF. Otherwise, the camera device auto-exposure related flash control (ON_AUTO_FLASH, ON_ALWAYS_FLASH, or ON_AUTO_FLASH_REDEYE) will override this control.

When set to OFF, the camera device will not fire flash for this capture.

When set to SINGLE, the camera device will fire flash regardless of the camera device's auto-exposure routine's result. When used in still capture case, this control should be used along with auto-exposure (AE) precapture metering sequence (android.control.aePrecaptureTrigger), otherwise, the image may be incorrectly exposed.

When set to TORCH, the flash will be on continuously. This mode can be used for use cases such as preview, auto-focus assist, still capture, or video recording.

The flash status will be reported by android.flash.state in the capture result metadata.

static
Property Name Type Description Units Range Tags
android.flash.info.available byte [public as boolean] [legacy]
  • FALSE
  • TRUE

Whether this camera device has a flash unit.

Details

Will be false if no flash is available.

If there is no flash unit, none of the flash controls do anything.

android.flash.info.chargeDuration int64 [system]

Time taken before flash can fire again

nanoseconds

0-1e9

Details

1 second too long/too short for recharge? Should this be power-dependent?

android.flash.colorTemperature byte [system]

The x,y whitepoint of the flash

pair of floats

0-1 for both

android.flash.maxEnergy byte [system]

Max energy output of the flash for a full power single flash

lumen-seconds

>= 0

dynamic
Property Name Type Description Units Range Tags
android.flash.firingPower byte [system]

Power for flash firing/torch

10 is max power; 0 is no flash. Linear

0 - 10

Details

Power for snapshot may use a different scale than for torch mode. Only one entry for torch mode will be used

android.flash.firingTime int64 [system]

Firing time of flash relative to start of exposure

nanoseconds

0-(exposure time-flash duration)

Details

Clamped to (0, exposure time - flash duration).

android.flash.mode byte [public] [legacy]
  • OFF

    Do not fire the flash for this capture.

  • SINGLE

    If the flash is available and charged, fire flash for this capture.

  • TORCH

    Transition flash to continuously on.

The desired mode for for the camera device's flash control.

Details

This control is only effective when flash unit is available (android.flash.info.available == true).

When this control is used, the android.control.aeMode must be set to ON or OFF. Otherwise, the camera device auto-exposure related flash control (ON_AUTO_FLASH, ON_ALWAYS_FLASH, or ON_AUTO_FLASH_REDEYE) will override this control.

When set to OFF, the camera device will not fire flash for this capture.

When set to SINGLE, the camera device will fire flash regardless of the camera device's auto-exposure routine's result. When used in still capture case, this control should be used along with auto-exposure (AE) precapture metering sequence (android.control.aePrecaptureTrigger), otherwise, the image may be incorrectly exposed.

When set to TORCH, the flash will be on continuously. This mode can be used for use cases such as preview, auto-focus assist, still capture, or video recording.

The flash status will be reported by android.flash.state in the capture result metadata.

android.flash.state byte [public] [limited]
  • UNAVAILABLE

    No flash on camera.

  • CHARGING

    Flash is charging and cannot be fired.

  • READY

    Flash is ready to fire.

  • FIRED

    Flash fired for this capture.

  • PARTIAL

    Flash partially illuminated this frame.

    This is usually due to the next or previous frame having the flash fire, and the flash spilling into this capture due to hardware limitations.

Current state of the flash unit.

Details

When the camera device doesn't have flash unit (i.e. android.flash.info.available == false), this state will always be UNAVAILABLE. Other states indicate the current flash status.

In certain conditions, this will be available on LEGACY devices:

In all other conditions the state will not be available on LEGACY devices (i.e. it will be null).

hotPixel
controls
Property Name Type Description Units Range Tags
android.hotPixel.mode byte [public]
  • OFF

    No hot pixel correction is applied.

    The frame rate must not be reduced relative to sensor raw output for this option.

    The hotpixel map may be returned in android.statistics.hotPixelMap.

  • FAST

    Hot pixel correction is applied, without reducing frame rate relative to sensor raw output.

    The hotpixel map may be returned in android.statistics.hotPixelMap.

  • HIGH_QUALITY

    High-quality hot pixel correction is applied, at a cost of possibly reduced frame rate relative to sensor raw output.

    The hotpixel map may be returned in android.statistics.hotPixelMap.

Operational mode for hot pixel correction.

android.hotPixel.availableHotPixelModes

Details

Hotpixel correction interpolates out, or otherwise removes, pixels that do not accurately measure the incoming light (i.e. pixels that are stuck at an arbitrary value or are oversensitive).

static
Property Name Type Description Units Range Tags
android.hotPixel.availableHotPixelModes byte x n [public as enumList]
list of enums

List of hot pixel correction modes for android.hotPixel.mode that are supported by this camera device.

Any value listed in android.hotPixel.mode

Details

FULL mode camera devices will always support FAST.

HAL Implementation Details

To avoid performance issues, there will be significantly fewer hot pixels than actual pixels on the camera sensor. HAL must support both FAST and HIGH_QUALITY if hot pixel correction control is available on the camera device, but the underlying implementation can be the same for both modes. That is, if the highest quality implementation on the camera device does not slow down capture rate, then FAST and HIGH_QUALITY will generate the same output.

dynamic
Property Name Type Description Units Range Tags
android.hotPixel.mode byte [public]
  • OFF

    No hot pixel correction is applied.

    The frame rate must not be reduced relative to sensor raw output for this option.

    The hotpixel map may be returned in android.statistics.hotPixelMap.

  • FAST

    Hot pixel correction is applied, without reducing frame rate relative to sensor raw output.

    The hotpixel map may be returned in android.statistics.hotPixelMap.

  • HIGH_QUALITY

    High-quality hot pixel correction is applied, at a cost of possibly reduced frame rate relative to sensor raw output.

    The hotpixel map may be returned in android.statistics.hotPixelMap.

Operational mode for hot pixel correction.

android.hotPixel.availableHotPixelModes

Details

Hotpixel correction interpolates out, or otherwise removes, pixels that do not accurately measure the incoming light (i.e. pixels that are stuck at an arbitrary value or are oversensitive).

jpeg
controls
Property Name Type Description Units Range Tags
android.jpeg.gpsLocation byte [public as location] [synthetic] [legacy]

A location object to use when generating image GPS metadata.

Details

Setting a location object in a request will include the GPS coordinates of the location into any JPEG images captured based on the request. These coordinates can then be viewed by anyone who receives the JPEG image.

android.jpeg.gpsCoordinates double x 3 [hidden] [legacy]
latitude, longitude, altitude. First two in degrees, the third in meters

GPS coordinates to include in output JPEG EXIF.

(-180 - 180], [-90,90], [-inf, inf]

android.jpeg.gpsProcessingMethod byte [hidden as string] [legacy]

32 characters describing GPS algorithm to include in EXIF.

UTF-8 null-terminated string
android.jpeg.gpsTimestamp int64 [hidden] [legacy]

Time GPS fix was made to include in EXIF.

UTC in seconds since January 1, 1970
android.jpeg.orientation int32 [public] [legacy]

The orientation for a JPEG image.

Degrees in multiples of 90

0, 90, 180, 270

Details

The clockwise rotation angle in degrees, relative to the orientation to the camera, that the JPEG picture needs to be rotated by, to be viewed upright.

Camera devices may either encode this value into the JPEG EXIF header, or rotate the image data to match this orientation. When the image data is rotated, the thumbnail data will also be rotated.

Note that this orientation is relative to the orientation of the camera sensor, given by android.sensor.orientation.

To translate from the device orientation given by the Android sensor APIs, the following sample code may be used:

private int getJpegOrientation(CameraCharacteristics c, int deviceOrientation) {
    if (deviceOrientation == android.view.OrientationEventListener.ORIENTATION_UNKNOWN) return 0;
    int sensorOrientation = c.get(CameraCharacteristics.SENSOR_ORIENTATION);

    // Round device orientation to a multiple of 90
    deviceOrientation = (deviceOrientation + 45) / 90 * 90;

    // Reverse device orientation for front-facing cameras
    boolean facingFront = c.get(CameraCharacteristics.LENS_FACING) == CameraCharacteristics.LENS_FACING_FRONT;
    if (facingFront) deviceOrientation = -deviceOrientation;

    // Calculate desired JPEG orientation relative to camera orientation to make
    // the image upright relative to the device orientation
    int jpegOrientation = (sensorOrientation + deviceOrientation + 360) % 360;

    return jpegOrientation;
}
android.jpeg.quality byte [public] [legacy]

Compression quality of the final JPEG image.

1-100; larger is higher quality

Details

85-95 is typical usage range.

android.jpeg.thumbnailQuality byte [public] [legacy]

Compression quality of JPEG thumbnail.

1-100; larger is higher quality

android.jpeg.thumbnailSize int32 x 2 [public as size] [legacy]

Resolution of embedded JPEG thumbnail.

android.jpeg.availableThumbnailSizes

Details

When set to (0, 0) value, the JPEG EXIF will not contain thumbnail, but the captured JPEG will still be a valid image.

For best results, when issuing a request for a JPEG image, the thumbnail size selected should have the same aspect ratio as the main JPEG output.

If the thumbnail image aspect ratio differs from the JPEG primary image aspect ratio, the camera device creates the thumbnail by cropping it from the primary image. For example, if the primary image has 4:3 aspect ratio, the thumbnail image has 16:9 aspect ratio, the primary image will be cropped vertically (letterbox) to generate the thumbnail image. The thumbnail image will always have a smaller Field Of View (FOV) than the primary image when aspect ratios differ.

When an android.jpeg.orientation of non-zero degree is requested, the camera device will handle thumbnail rotation in one of the following ways:

  • Set the EXIF orientation flag and keep jpeg and thumbnail image data unrotated.
  • Rotate the jpeg and thumbnail image data and not set EXIF orientation flag. In this case, LIMITED or FULL hardware level devices will report rotated thumnail size in capture result, so the width and height will be interchanged if 90 or 270 degree orientation is requested. LEGACY device will always report unrotated thumbnail size.
HAL Implementation Details

The HAL must not squeeze or stretch the downscaled primary image to generate thumbnail. The cropping must be done on the primary jpeg image rather than the sensor active array. The stream cropping rule specified by "S5. Cropping" in camera3.h doesn't apply to the thumbnail image cropping.

static
Property Name Type Description Units Range Tags
android.jpeg.availableThumbnailSizes int32 x 2 x n [public as size] [legacy]

List of JPEG thumbnail sizes for android.jpeg.thumbnailSize supported by this camera device.

Details

This list will include at least one non-zero resolution, plus (0,0) for indicating no thumbnail should be generated.

Below condiditions will be satisfied for this size list:

  • The sizes will be sorted by increasing pixel area (width x height). If several resolutions have the same area, they will be sorted by increasing width.
  • The aspect ratio of the largest thumbnail size will be same as the aspect ratio of largest JPEG output size in android.scaler.availableStreamConfigurations. The largest size is defined as the size that has the largest pixel area in a given size list.
  • Each output JPEG size in android.scaler.availableStreamConfigurations will have at least one corresponding size that has the same aspect ratio in availableThumbnailSizes, and vice versa.
  • All non-(0, 0) sizes will have non-zero widths and heights.
android.jpeg.maxSize int32 [system]

Maximum size in bytes for the compressed JPEG buffer

Must be large enough to fit any JPEG produced by the camera

Details

This is used for sizing the gralloc buffers for JPEG

dynamic
Property Name Type Description Units Range Tags
android.jpeg.gpsLocation byte [public as location] [synthetic] [legacy]

A location object to use when generating image GPS metadata.

Details

Setting a location object in a request will include the GPS coordinates of the location into any JPEG images captured based on the request. These coordinates can then be viewed by anyone who receives the JPEG image.

android.jpeg.gpsCoordinates double x 3 [hidden] [legacy]
latitude, longitude, altitude. First two in degrees, the third in meters

GPS coordinates to include in output JPEG EXIF.

(-180 - 180], [-90,90], [-inf, inf]

android.jpeg.gpsProcessingMethod byte [hidden as string] [legacy]

32 characters describing GPS algorithm to include in EXIF.

UTF-8 null-terminated string
android.jpeg.gpsTimestamp int64 [hidden] [legacy]

Time GPS fix was made to include in EXIF.

UTC in seconds since January 1, 1970
android.jpeg.orientation int32 [public] [legacy]

The orientation for a JPEG image.

Degrees in multiples of 90

0, 90, 180, 270

Details

The clockwise rotation angle in degrees, relative to the orientation to the camera, that the JPEG picture needs to be rotated by, to be viewed upright.

Camera devices may either encode this value into the JPEG EXIF header, or rotate the image data to match this orientation. When the image data is rotated, the thumbnail data will also be rotated.

Note that this orientation is relative to the orientation of the camera sensor, given by android.sensor.orientation.

To translate from the device orientation given by the Android sensor APIs, the following sample code may be used:

private int getJpegOrientation(CameraCharacteristics c, int deviceOrientation) {
    if (deviceOrientation == android.view.OrientationEventListener.ORIENTATION_UNKNOWN) return 0;
    int sensorOrientation = c.get(CameraCharacteristics.SENSOR_ORIENTATION);

    // Round device orientation to a multiple of 90
    deviceOrientation = (deviceOrientation + 45) / 90 * 90;

    // Reverse device orientation for front-facing cameras
    boolean facingFront = c.get(CameraCharacteristics.LENS_FACING) == CameraCharacteristics.LENS_FACING_FRONT;
    if (facingFront) deviceOrientation = -deviceOrientation;

    // Calculate desired JPEG orientation relative to camera orientation to make
    // the image upright relative to the device orientation
    int jpegOrientation = (sensorOrientation + deviceOrientation + 360) % 360;

    return jpegOrientation;
}
android.jpeg.quality byte [public] [legacy]

Compression quality of the final JPEG image.

1-100; larger is higher quality

Details

85-95 is typical usage range.

android.jpeg.size int32 [system]

The size of the compressed JPEG image, in bytes

>= 0

Details

If no JPEG output is produced for the request, this must be 0.

Otherwise, this describes the real size of the compressed JPEG image placed in the output stream. More specifically, if android.jpeg.maxSize = 1000000, and a specific capture has android.jpeg.size = 500000, then the output buffer from the JPEG stream will be 1000000 bytes, of which the first 500000 make up the real data.

android.jpeg.thumbnailQuality byte [public] [legacy]

Compression quality of JPEG thumbnail.

1-100; larger is higher quality

android.jpeg.thumbnailSize int32 x 2 [public as size] [legacy]

Resolution of embedded JPEG thumbnail.

android.jpeg.availableThumbnailSizes

Details

When set to (0, 0) value, the JPEG EXIF will not contain thumbnail, but the captured JPEG will still be a valid image.

For best results, when issuing a request for a JPEG image, the thumbnail size selected should have the same aspect ratio as the main JPEG output.

If the thumbnail image aspect ratio differs from the JPEG primary image aspect ratio, the camera device creates the thumbnail by cropping it from the primary image. For example, if the primary image has 4:3 aspect ratio, the thumbnail image has 16:9 aspect ratio, the primary image will be cropped vertically (letterbox) to generate the thumbnail image. The thumbnail image will always have a smaller Field Of View (FOV) than the primary image when aspect ratios differ.

When an android.jpeg.orientation of non-zero degree is requested, the camera device will handle thumbnail rotation in one of the following ways:

  • Set the EXIF orientation flag and keep jpeg and thumbnail image data unrotated.
  • Rotate the jpeg and thumbnail image data and not set EXIF orientation flag. In this case, LIMITED or FULL hardware level devices will report rotated thumnail size in capture result, so the width and height will be interchanged if 90 or 270 degree orientation is requested. LEGACY device will always report unrotated thumbnail size.
HAL Implementation Details

The HAL must not squeeze or stretch the downscaled primary image to generate thumbnail. The cropping must be done on the primary jpeg image rather than the sensor active array. The stream cropping rule specified by "S5. Cropping" in camera3.h doesn't apply to the thumbnail image cropping.

lens
controls
Property Name Type Description Units Range Tags
android.lens.aperture float [public] [full]

The desired lens aperture size, as a ratio of lens focal length to the effective aperture diameter.

The f-number (f/N)

android.lens.info.availableApertures

Details

Setting this value is only supported on the camera devices that have a variable aperture lens.

When this is supported and android.control.aeMode is OFF, this can be set along with android.sensor.exposureTime, android.sensor.sensitivity, and android.sensor.frameDuration to achieve manual exposure control.

The requested aperture value may take several frames to reach the requested value; the camera device will report the current (intermediate) aperture size in capture result metadata while the aperture is changing. While the aperture is still changing, android.lens.state will be set to MOVING.

When this is supported and android.control.aeMode is one of the ON modes, this will be overridden by the camera device auto-exposure algorithm, the overridden values are then provided back to the user in the corresponding result.

android.lens.filterDensity float [public] [full]

The desired setting for the lens neutral density filter(s).

Exposure Value (EV)

android.lens.info.availableFilterDensities

Details

This control will not be supported on most camera devices.

Lens filters are typically used to lower the amount of light the sensor is exposed to (measured in steps of EV). As used here, an EV step is the standard logarithmic representation, which are non-negative, and inversely proportional to the amount of light hitting the sensor. For example, setting this to 0 would result in no reduction of the incoming light, and setting this to 2 would mean that the filter is set to reduce incoming light by two stops (allowing 1/4 of the prior amount of light to the sensor).

It may take several frames before the lens filter density changes to the requested value. While the filter density is still changing, android.lens.state will be set to MOVING.

android.lens.focalLength float [public] [legacy]

The desired lens focal length; used for optical zoom.

Millimeters

android.lens.info.availableFocalLengths

Details

This setting controls the physical focal length of the camera device's lens. Changing the focal length changes the field of view of the camera device, and is usually used for optical zoom.

Like android.lens.focusDistance and android.lens.aperture, this setting won't be applied instantaneously, and it may take several frames before the lens can change to the requested focal length. While the focal length is still changing, android.lens.state will be set to MOVING.

Optical zoom will not be supported on most devices.

android.lens.focusDistance float [public] [full]

Desired distance to plane of sharpest focus, measured from frontmost surface of the lens.

See android.lens.info.focusDistanceCalibration for details

>= 0

Details

This control can be used for setting manual focus, on devices that support the MANUAL_SENSOR capability and have a variable-focus lens (see android.lens.info.minimumFocusDistance).

A value of 0.0f means infinity focus. The value set will be clamped to [0.0f, android.lens.info.minimumFocusDistance].

Like android.lens.focalLength, this setting won't be applied instantaneously, and it may take several frames before the lens can move to the requested focus distance. While the lens is still moving, android.lens.state will be set to MOVING.

LEGACY devices support at most setting this to 0.0f for infinity focus.

android.lens.opticalStabilizationMode byte [public] [limited]
  • OFF

    Optical stabilization is unavailable.

  • ON [optional]

    Optical stabilization is enabled.

Sets whether the camera device uses optical image stabilization (OIS) when capturing images.

android.lens.info.availableOpticalStabilization

Details

OIS is used to compensate for motion blur due to small movements of the camera during capture. Unlike digital image stabilization (android.control.videoStabilizationMode), OIS makes use of mechanical elements to stabilize the camera sensor, and thus allows for longer exposure times before camera shake becomes apparent.

Switching between different optical stabilization modes may take several frames to initialize, the camera device will report the current mode in capture result metadata. For example, When "ON" mode is requested, the optical stabilization modes in the first several capture results may still be "OFF", and it will become "ON" when the initialization is done.

If a camera device supports both OIS and digital image stabilization (android.control.videoStabilizationMode), turning both modes on may produce undesirable interaction, so it is recommended not to enable both at the same time.

Not all devices will support OIS; see android.lens.info.availableOpticalStabilization for available controls.

static
Property Name Type Description Units Range Tags
android.lens.info.availableApertures float x n [public] [full]

List of aperture size values for android.lens.aperture that are supported by this camera device.

The aperture f-number
Details

If the camera device doesn't support a variable lens aperture, this list will contain only one value, which is the fixed aperture size.

If the camera device supports a variable aperture, the aperture values in this list will be sorted in ascending order.

android.lens.info.availableFilterDensities float x n [public] [full]

List of neutral density filter values for android.lens.filterDensity that are supported by this camera device.

Exposure value (EV)

Values are >= 0

Details

If a neutral density filter is not supported by this camera device, this list will contain only 0. Otherwise, this list will include every filter density supported by the camera device, in ascending order.

android.lens.info.availableFocalLengths float x n [public] [legacy]
The list of available focal lengths

List of focal lengths for android.lens.focalLength that are supported by this camera device.

Millimeters

Values are > 0

Details

If optical zoom is not supported, this list will only contain a single value corresponding to the fixed focal length of the device. Otherwise, this list will include every focal length supported by the camera device, in ascending order.

android.lens.info.availableOpticalStabilization byte x n [public as enumList] [limited]
list of enums

List of optical image stabilization (OIS) modes for android.lens.opticalStabilizationMode that are supported by this camera device.

Any value listed in android.lens.opticalStabilizationMode

Details

If OIS is not supported by a given camera device, this list will contain only OFF.

android.lens.info.hyperfocalDistance float [public] [limited]

Hyperfocal distance for this lens.

See android.lens.info.focusDistanceCalibration for details

If lens is fixed focus, >= 0. If lens has focuser unit, the value is within (0.0f, android.lens.info.minimumFocusDistance]

Details

If the lens is not fixed focus, the camera device will report this field when android.lens.info.focusDistanceCalibration is APPROXIMATE or CALIBRATED.

android.lens.info.minimumFocusDistance float [public] [limited]

Shortest distance from frontmost surface of the lens that can be brought into sharp focus.

See android.lens.info.focusDistanceCalibration for details

>= 0

Details

If the lens is fixed-focus, this will be 0.

HAL Implementation Details

Mandatory for FULL devices; LIMITED devices must always set this value to 0 for fixed-focus; and may omit the minimum focus distance otherwise.

This field is also mandatory for all devices advertising the MANUAL_SENSOR capability.

android.lens.info.shadingMapSize int32 x 2 [hidden as size] [full]
width and height (N, M) of lens shading map provided by the camera device.

Dimensions of lens shading map.

Both values >= 1

Details

The map should be on the order of 30-40 rows and columns, and must be smaller than 64x64.

android.lens.info.focusDistanceCalibration byte [public] [limited]
  • UNCALIBRATED

    The lens focus distance is not accurate, and the units used for android.lens.focusDistance do not correspond to any physical units.

    Setting the lens to the same focus distance on separate occasions may result in a different real focus distance, depending on factors such as the orientation of the device, the age of the focusing mechanism, and the device temperature. The focus distance value will still be in the range of [0, android.lens.info.minimumFocusDistance], where 0 represents the farthest focus.

  • APPROXIMATE

    The lens focus distance is measured in diopters.

    However, setting the lens to the same focus distance on separate occasions may result in a different real focus distance, depending on factors such as the orientation of the device, the age of the focusing mechanism, and the device temperature.

  • CALIBRATED

    The lens focus distance is measured in diopters, and is calibrated.

    The lens mechanism is calibrated so that setting the same focus distance is repeatable on multiple occasions with good accuracy, and the focus distance corresponds to the real physical distance to the plane of best focus.

The lens focus distance calibration quality.

Details

The lens focus distance calibration quality determines the reliability of focus related metadata entries, i.e. android.lens.focusDistance, android.lens.focusRange, android.lens.info.hyperfocalDistance, and android.lens.info.minimumFocusDistance.

APPROXIMATE and CALIBRATED devices report the focus metadata in units of diopters (1/meter), so 0.0f represents focusing at infinity, and increasing positive numbers represent focusing closer and closer to the camera device. The focus distance control also uses diopters on these devices.

UNCALIBRATED devices do not use units that are directly comparable to any real physical measurement, but 0.0f still represents farthest focus, and android.lens.info.minimumFocusDistance represents the nearest focus the device can achieve.

HAL Implementation Details

For devices advertise APPROXIMATE quality or higher, diopters 0 (infinity focus) must work. When autofocus is disabled (android.control.afMode == OFF) and the lens focus distance is set to 0 diopters (android.lens.focusDistance == 0), the lens will move to focus at infinity and is stably focused at infinity even if the device tilts. It may take the lens some time to move; during the move the lens state should be MOVING and the output diopter value should be changing toward 0.

android.lens.facing byte [public] [legacy]
  • FRONT

    The camera device faces the same direction as the device's screen.

  • BACK

    The camera device faces the opposite direction as the device's screen.

  • EXTERNAL

    The camera device is an external camera, and has no fixed facing relative to the device's screen.

Direction the camera faces relative to device screen.

android.lens.poseRotation float x 4 [public]

The orientation of the camera relative to the sensor coordinate system.

Quaternion coefficients
Details

The four coefficients that describe the quaternion rotation from the Android sensor coordinate system to a camera-aligned coordinate system where the X-axis is aligned with the long side of the image sensor, the Y-axis is aligned with the short side of the image sensor, and the Z-axis is aligned with the optical axis of the sensor.

To convert from the quaternion coefficients (x,y,z,w) to the axis of rotation (a_x, a_y, a_z) and rotation amount theta, the following formulas can be used:

 theta = 2 * acos(w)
a_x = x / sin(theta/2)
a_y = y / sin(theta/2)
a_z = z / sin(theta/2)

To create a 3x3 rotation matrix that applies the rotation defined by this quaternion, the following matrix can be used:

R = [ 1 - 2y^2 - 2z^2,       2xy - 2zw,       2xz + 2yw,
           2xy + 2zw, 1 - 2x^2 - 2z^2,       2yz - 2xw,
           2xz - 2yw,       2yz + 2xw, 1 - 2x^2 - 2y^2 ]

This matrix can then be used to apply the rotation to a column vector point with

p' = Rp

where p is in the device sensor coordinate system, and p' is in the camera-oriented coordinate system.

android.lens.poseTranslation float x 3 [public]

Position of the camera optical center.

Meters
Details

The position of the camera device's lens optical center, as a three-dimensional vector (x,y,z), relative to the optical center of the largest camera device facing in the same direction as this camera, in the Android sensor coordinate axes. Note that only the axis definitions are shared with the sensor coordinate system, but not the origin.

If this device is the largest or only camera device with a given facing, then this position will be (0, 0, 0); a camera device with a lens optical center located 3 cm from the main sensor along the +X axis (to the right from the user's perspective) will report (0.03, 0, 0).

To transform a pixel coordinates between two cameras facing the same direction, first the source camera android.lens.radialDistortion must be corrected for. Then the source camera android.lens.intrinsicCalibration needs to be applied, followed by the android.lens.poseRotation of the source camera, the translation of the source camera relative to the destination camera, the android.lens.poseRotation of the destination camera, and finally the inverse of android.lens.intrinsicCalibration of the destination camera. This obtains a radial-distortion-free coordinate in the destination camera pixel coordinates.

To compare this against a real image from the destination camera, the destination camera image then needs to be corrected for radial distortion before comparison or sampling.

android.lens.intrinsicCalibration float x 5 [public]

The parameters for this camera device's intrinsic calibration.

Pixels in the android.sensor.info.preCorrectionActiveArraySize coordinate system.
Details

The five calibration parameters that describe the transform from camera-centric 3D coordinates to sensor pixel coordinates:

[f_x, f_y, c_x, c_y, s]

Where f_x and f_y are the horizontal and vertical focal lengths, [c_x, c_y] is the position of the optical axis, and s is a skew parameter for the sensor plane not being aligned with the lens plane.

These are typically used within a transformation matrix K:

K = [ f_x,   s, c_x,
       0, f_y, c_y,
       0    0,   1 ]

which can then be combined with the camera pose rotation R and translation t (android.lens.poseRotation and android.lens.poseTranslation, respective) to calculate the complete transform from world coordinates to pixel coordinates:

P = [ K 0   * [ R t
     0 1 ]     0 1 ]

and with p_w being a point in the world coordinate system and p_s being a point in the camera active pixel array coordinate system, and with the mapping including the homogeneous division by z:

 p_h = (x_h, y_h, z_h) = P p_w
p_s = p_h / z_h

so [x_s, y_s] is the pixel coordinates of the world point, z_s = 1, and w_s is a measurement of disparity (depth) in pixel coordinates.

Note that the coordinate system for this transform is the android.sensor.info.preCorrectionActiveArraySize system, where (0,0) is the top-left of the preCorrectionActiveArraySize rectangle. Once the pose and intrinsic calibration transforms have been applied to a world point, then the android.lens.radialDistortion transform needs to be applied, and the result adjusted to be in the android.sensor.info.activeArraySize coordinate system (where (0, 0) is the top-left of the activeArraySize rectangle), to determine the final pixel coordinate of the world point for processed (non-RAW) output buffers.

android.lens.radialDistortion float x 6 [public]

The correction coefficients to correct for this camera device's radial and tangential lens distortion.

Unitless coefficients.
Details

Four radial distortion coefficients [kappa_0, kappa_1, kappa_2, kappa_3] and two tangential distortion coefficients [kappa_4, kappa_5] that can be used to correct the lens's geometric distortion with the mapping equations:

 x_c = x_i * ( kappa_0 + kappa_1 * r^2 + kappa_2 * r^4 + kappa_3 * r^6 ) +
       kappa_4 * (2 * x_i * y_i) + kappa_5 * ( r^2 + 2 * x_i^2 )
 y_c = y_i * ( kappa_0 + kappa_1 * r^2 + kappa_2 * r^4 + kappa_3 * r^6 ) +
       kappa_5 * (2 * x_i * y_i) + kappa_4 * ( r^2 + 2 * y_i^2 )

Here, [x_c, y_c] are the coordinates to sample in the input image that correspond to the pixel values in the corrected image at the coordinate [x_i, y_i]:

 correctedImage(x_i, y_i) = sample_at(x_c, y_c, inputImage)

The pixel coordinates are defined in a normalized coordinate system related to the android.lens.intrinsicCalibration calibration fields. Both [x_i, y_i] and [x_c, y_c] have (0,0) at the lens optical center [c_x, c_y]. The maximum magnitudes of both x and y coordinates are normalized to be 1 at the edge further from the optical center, so the range for both dimensions is -1 <= x <= 1.

Finally, r represents the radial distance from the optical center, r^2 = x_i^2 + y_i^2, and its magnitude is therefore no larger than |r| <= sqrt(2).

The distortion model used is the Brown-Conrady model.

dynamic
Property Name Type Description Units Range Tags
android.lens.aperture float [public] [full]

The desired lens aperture size, as a ratio of lens focal length to the effective aperture diameter.

The f-number (f/N)

android.lens.info.availableApertures

Details

Setting this value is only supported on the camera devices that have a variable aperture lens.

When this is supported and android.control.aeMode is OFF, this can be set along with android.sensor.exposureTime, android.sensor.sensitivity, and android.sensor.frameDuration to achieve manual exposure control.

The requested aperture value may take several frames to reach the requested value; the camera device will report the current (intermediate) aperture size in capture result metadata while the aperture is changing. While the aperture is still changing, android.lens.state will be set to MOVING.

When this is supported and android.control.aeMode is one of the ON modes, this will be overridden by the camera device auto-exposure algorithm, the overridden values are then provided back to the user in the corresponding result.

android.lens.filterDensity float [public] [full]

The desired setting for the lens neutral density filter(s).

Exposure Value (EV)

android.lens.info.availableFilterDensities

Details

This control will not be supported on most camera devices.

Lens filters are typically used to lower the amount of light the sensor is exposed to (measured in steps of EV). As used here, an EV step is the standard logarithmic representation, which are non-negative, and inversely proportional to the amount of light hitting the sensor. For example, setting this to 0 would result in no reduction of the incoming light, and setting this to 2 would mean that the filter is set to reduce incoming light by two stops (allowing 1/4 of the prior amount of light to the sensor).

It may take several frames before the lens filter density changes to the requested value. While the filter density is still changing, android.lens.state will be set to MOVING.

android.lens.focalLength float [public] [legacy]

The desired lens focal length; used for optical zoom.

Millimeters

android.lens.info.availableFocalLengths

Details

This setting controls the physical focal length of the camera device's lens. Changing the focal length changes the field of view of the camera device, and is usually used for optical zoom.

Like android.lens.focusDistance and android.lens.aperture, this setting won't be applied instantaneously, and it may take several frames before the lens can change to the requested focal length. While the focal length is still changing, android.lens.state will be set to MOVING.

Optical zoom will not be supported on most devices.

android.lens.focusDistance float [public] [full]

Desired distance to plane of sharpest focus, measured from frontmost surface of the lens.

See android.lens.info.focusDistanceCalibration for details

>= 0

Details

Should be zero for fixed-focus cameras

android.lens.focusRange float x 2 [public as pairFloatFloat] [limited]
Range of scene distances that are in focus

The range of scene distances that are in sharp focus (depth of field).

A pair of focus distances in diopters: (near, far); see android.lens.info.focusDistanceCalibration for details.

>=0

Details

If variable focus not supported, can still report fixed depth of field range

android.lens.opticalStabilizationMode byte [public] [limited]
  • OFF

    Optical stabilization is unavailable.

  • ON [optional]

    Optical stabilization is enabled.

Sets whether the camera device uses optical image stabilization (OIS) when capturing images.

android.lens.info.availableOpticalStabilization

Details

OIS is used to compensate for motion blur due to small movements of the camera during capture. Unlike digital image stabilization (android.control.videoStabilizationMode), OIS makes use of mechanical elements to stabilize the camera sensor, and thus allows for longer exposure times before camera shake becomes apparent.

Switching between different optical stabilization modes may take several frames to initialize, the camera device will report the current mode in capture result metadata. For example, When "ON" mode is requested, the optical stabilization modes in the first several capture results may still be "OFF", and it will become "ON" when the initialization is done.

If a camera device supports both OIS and digital image stabilization (android.control.videoStabilizationMode), turning both modes on may produce undesirable interaction, so it is recommended not to enable both at the same time.

Not all devices will support OIS; see android.lens.info.availableOpticalStabilization for available controls.

android.lens.state byte [public] [limited]

Current lens status.

Details

For lens parameters android.lens.focalLength, android.lens.focusDistance, android.lens.filterDensity and android.lens.aperture, when changes are requested, they may take several frames to reach the requested values. This state indicates the current status of the lens parameters.

When the state is STATIONARY, the lens parameters are not changing. This could be either because the parameters are all fixed, or because the lens has had enough time to reach the most recently-requested values. If all these lens parameters are not changable for a camera device, as listed below:

Then this state will always be STATIONARY.

When the state is MOVING, it indicates that at least one of the lens parameters is changing.

android.lens.poseRotation float x 4 [public]

The orientation of the camera relative to the sensor coordinate system.

Quaternion coefficients
Details

The four coefficients that describe the quaternion rotation from the Android sensor coordinate system to a camera-aligned coordinate system where the X-axis is aligned with the long side of the image sensor, the Y-axis is aligned with the short side of the image sensor, and the Z-axis is aligned with the optical axis of the sensor.

To convert from the quaternion coefficients (x,y,z,w) to the axis of rotation (a_x, a_y, a_z) and rotation amount theta, the following formulas can be used:

 theta = 2 * acos(w)
a_x = x / sin(theta/2)
a_y = y / sin(theta/2)
a_z = z / sin(theta/2)

To create a 3x3 rotation matrix that applies the rotation defined by this quaternion, the following matrix can be used:

R = [ 1 - 2y^2 - 2z^2,       2xy - 2zw,       2xz + 2yw,
           2xy + 2zw, 1 - 2x^2 - 2z^2,       2yz - 2xw,
           2xz - 2yw,       2yz + 2xw, 1 - 2x^2 - 2y^2 ]

This matrix can then be used to apply the rotation to a column vector point with

p' = Rp

where p is in the device sensor coordinate system, and p' is in the camera-oriented coordinate system.

android.lens.poseTranslation float x 3 [public]

Position of the camera optical center.

Meters
Details

The position of the camera device's lens optical center, as a three-dimensional vector (x,y,z), relative to the optical center of the largest camera device facing in the same direction as this camera, in the Android sensor coordinate axes. Note that only the axis definitions are shared with the sensor coordinate system, but not the origin.

If this device is the largest or only camera device with a given facing, then this position will be (0, 0, 0); a camera device with a lens optical center located 3 cm from the main sensor along the +X axis (to the right from the user's perspective) will report (0.03, 0, 0).

To transform a pixel coordinates between two cameras facing the same direction, first the source camera android.lens.radialDistortion must be corrected for. Then the source camera android.lens.intrinsicCalibration needs to be applied, followed by the android.lens.poseRotation of the source camera, the translation of the source camera relative to the destination camera, the android.lens.poseRotation of the destination camera, and finally the inverse of android.lens.intrinsicCalibration of the destination camera. This obtains a radial-distortion-free coordinate in the destination camera pixel coordinates.

To compare this against a real image from the destination camera, the destination camera image then needs to be corrected for radial distortion before comparison or sampling.

android.lens.intrinsicCalibration float x 5 [public]

The parameters for this camera device's intrinsic calibration.

Pixels in the android.sensor.info.preCorrectionActiveArraySize coordinate system.
Details

The five calibration parameters that describe the transform from camera-centric 3D coordinates to sensor pixel coordinates:

[f_x, f_y, c_x, c_y, s]

Where f_x and f_y are the horizontal and vertical focal lengths, [c_x, c_y] is the position of the optical axis, and s is a skew parameter for the sensor plane not being aligned with the lens plane.

These are typically used within a transformation matrix K:

K = [ f_x,   s, c_x,
       0, f_y, c_y,
       0    0,   1 ]

which can then be combined with the camera pose rotation R and translation t (android.lens.poseRotation and android.lens.poseTranslation, respective) to calculate the complete transform from world coordinates to pixel coordinates:

P = [ K 0   * [ R t
     0 1 ]     0 1 ]

and with p_w being a point in the world coordinate system and p_s being a point in the camera active pixel array coordinate system, and with the mapping including the homogeneous division by z:

 p_h = (x_h, y_h, z_h) = P p_w
p_s = p_h / z_h

so [x_s, y_s] is the pixel coordinates of the world point, z_s = 1, and w_s is a measurement of disparity (depth) in pixel coordinates.

Note that the coordinate system for this transform is the android.sensor.info.preCorrectionActiveArraySize system, where (0,0) is the top-left of the preCorrectionActiveArraySize rectangle. Once the pose and intrinsic calibration transforms have been applied to a world point, then the android.lens.radialDistortion transform needs to be applied, and the result adjusted to be in the android.sensor.info.activeArraySize coordinate system (where (0, 0) is the top-left of the activeArraySize rectangle), to determine the final pixel coordinate of the world point for processed (non-RAW) output buffers.

android.lens.radialDistortion float x 6 [public]

The correction coefficients to correct for this camera device's radial and tangential lens distortion.

Unitless coefficients.
Details

Four radial distortion coefficients [kappa_0, kappa_1, kappa_2, kappa_3] and two tangential distortion coefficients [kappa_4, kappa_5] that can be used to correct the lens's geometric distortion with the mapping equations:

 x_c = x_i * ( kappa_0 + kappa_1 * r^2 + kappa_2 * r^4 + kappa_3 * r^6 ) +
       kappa_4 * (2 * x_i * y_i) + kappa_5 * ( r^2 + 2 * x_i^2 )
 y_c = y_i * ( kappa_0 + kappa_1 * r^2 + kappa_2 * r^4 + kappa_3 * r^6 ) +
       kappa_5 * (2 * x_i * y_i) + kappa_4 * ( r^2 + 2 * y_i^2 )

Here, [x_c, y_c] are the coordinates to sample in the input image that correspond to the pixel values in the corrected image at the coordinate [x_i, y_i]:

 correctedImage(x_i, y_i) = sample_at(x_c, y_c, inputImage)

The pixel coordinates are defined in a normalized coordinate system related to the android.lens.intrinsicCalibration calibration fields. Both [x_i, y_i] and [x_c, y_c] have (0,0) at the lens optical center [c_x, c_y]. The maximum magnitudes of both x and y coordinates are normalized to be 1 at the edge further from the optical center, so the range for both dimensions is -1 <= x <= 1.

Finally, r represents the radial distance from the optical center, r^2 = x_i^2 + y_i^2, and its magnitude is therefore no larger than |r| <= sqrt(2).

The distortion model used is the Brown-Conrady model.

noiseReduction
controls
Property Name Type Description Units Range Tags
android.noiseReduction.mode byte [public] [full]
  • OFF

    No noise reduction is applied.

  • FAST

    Noise reduction is applied without reducing frame rate relative to sensor output. It may be the same as OFF if noise reduction will reduce frame rate relative to sensor.

  • HIGH_QUALITY

    High-quality noise reduction is applied, at the cost of possibly reduced frame rate relative to sensor output.

  • MINIMAL [optional]

    MINIMAL noise reduction is applied without reducing frame rate relative to sensor output.

  • ZERO_SHUTTER_LAG [optional]

    Noise reduction is applied at different levels for different output streams, based on resolution. Streams at maximum recording resolution (see CameraDevice#createCaptureSession) or below have noise reduction applied, while higher-resolution streams have MINIMAL (if supported) or no noise reduction applied (if MINIMAL is not supported.) The degree of noise reduction for low-resolution streams is tuned so that frame rate is not impacted, and the quality is equal to or better than FAST (since it is only applied to lower-resolution outputs, quality may improve from FAST).

    This mode is intended to be used by applications operating in a zero-shutter-lag mode with YUV or PRIVATE reprocessing, where the application continuously captures high-resolution intermediate buffers into a circular buffer, from which a final image is produced via reprocessing when a user takes a picture. For such a use case, the high-resolution buffers must not have noise reduction applied to maximize efficiency of preview and to avoid over-applying noise filtering when reprocessing, while low-resolution buffers (used for recording or preview, generally) need noise reduction applied for reasonable preview quality.

    This mode is guaranteed to be supported by devices that support either the YUV_REPROCESSING or PRIVATE_REPROCESSING capabilities (android.request.availableCapabilities lists either of those capabilities) and it will be the default mode for CAMERA3_TEMPLATE_ZERO_SHUTTER_LAG template.

Mode of operation for the noise reduction algorithm.

android.noiseReduction.availableNoiseReductionModes

Details

The noise reduction algorithm attempts to improve image quality by removing excessive noise added by the capture process, especially in dark conditions.

OFF means no noise reduction will be applied by the camera device, for both raw and YUV domain.

MINIMAL means that only sensor raw domain basic noise reduction is enabled ,to remove demosaicing or other processing artifacts. For YUV_REPROCESSING, MINIMAL is same as OFF. This mode is optional, may not be support by all devices. The application should check android.noiseReduction.availableNoiseReductionModes before using it.

FAST/HIGH_QUALITY both mean camera device determined noise filtering will be applied. HIGH_QUALITY mode indicates that the camera device will use the highest-quality noise filtering algorithms, even if it slows down capture rate. FAST means the camera device will not slow down capture rate when applying noise filtering. FAST may be the same as MINIMAL if MINIMAL is listed, or the same as OFF if any noise filtering will slow down capture rate. Every output stream will have a similar amount of enhancement applied.

ZERO_SHUTTER_LAG is meant to be used by applications that maintain a continuous circular buffer of high-resolution images during preview and reprocess image(s) from that buffer into a final capture when triggered by the user. In this mode, the camera device applies noise reduction to low-resolution streams (below maximum recording resolution) to maximize preview quality, but does not apply noise reduction to high-resolution streams, since those will be reprocessed later if necessary.

For YUV_REPROCESSING, these FAST/HIGH_QUALITY modes both mean that the camera device will apply FAST/HIGH_QUALITY YUV domain noise reduction, respectively. The camera device may adjust the noise reduction parameters for best image quality based on the android.reprocess.effectiveExposureFactor if it is set.

HAL Implementation Details

For YUV_REPROCESSING The HAL can use android.reprocess.effectiveExposureFactor to adjust the internal noise reduction parameters appropriately to get the best quality images.

android.noiseReduction.strength byte [system]

Control the amount of noise reduction applied to the images

1-10; 10 is max noise reduction

1 - 10

static
Property Name Type Description Units Range Tags
android.noiseReduction.availableNoiseReductionModes byte x n [public as enumList] [limited]
list of enums

List of noise reduction modes for android.noiseReduction.mode that are supported by this camera device.

Any value listed in android.noiseReduction.mode

Details

Full-capability camera devices will always support OFF and FAST.

Camera devices that support YUV_REPROCESSING or PRIVATE_REPROCESSING will support ZERO_SHUTTER_LAG.

Legacy-capability camera devices will only support FAST mode.

HAL Implementation Details

HAL must support both FAST and HIGH_QUALITY if noise reduction control is available on the camera device, but the underlying implementation can be the same for both modes. That is, if the highest quality implementation on the camera device does not slow down capture rate, then FAST and HIGH_QUALITY will generate the same output.

dynamic
Property Name Type Description Units Range Tags
android.noiseReduction.mode byte [public] [full]
  • OFF

    No noise reduction is applied.

  • FAST

    Noise reduction is applied without reducing frame rate relative to sensor output. It may be the same as OFF if noise reduction will reduce frame rate relative to sensor.

  • HIGH_QUALITY

    High-quality noise reduction is applied, at the cost of possibly reduced frame rate relative to sensor output.

  • MINIMAL [optional]

    MINIMAL noise reduction is applied without reducing frame rate relative to sensor output.

  • ZERO_SHUTTER_LAG [optional]

    Noise reduction is applied at different levels for different output streams, based on resolution. Streams at maximum recording resolution (see CameraDevice#createCaptureSession) or below have noise reduction applied, while higher-resolution streams have MINIMAL (if supported) or no noise reduction applied (if MINIMAL is not supported.) The degree of noise reduction for low-resolution streams is tuned so that frame rate is not impacted, and the quality is equal to or better than FAST (since it is only applied to lower-resolution outputs, quality may improve from FAST).

    This mode is intended to be used by applications operating in a zero-shutter-lag mode with YUV or PRIVATE reprocessing, where the application continuously captures high-resolution intermediate buffers into a circular buffer, from which a final image is produced via reprocessing when a user takes a picture. For such a use case, the high-resolution buffers must not have noise reduction applied to maximize efficiency of preview and to avoid over-applying noise filtering when reprocessing, while low-resolution buffers (used for recording or preview, generally) need noise reduction applied for reasonable preview quality.

    This mode is guaranteed to be supported by devices that support either the YUV_REPROCESSING or PRIVATE_REPROCESSING capabilities (android.request.availableCapabilities lists either of those capabilities) and it will be the default mode for CAMERA3_TEMPLATE_ZERO_SHUTTER_LAG template.

Mode of operation for the noise reduction algorithm.

android.noiseReduction.availableNoiseReductionModes

Details

The noise reduction algorithm attempts to improve image quality by removing excessive noise added by the capture process, especially in dark conditions.

OFF means no noise reduction will be applied by the camera device, for both raw and YUV domain.

MINIMAL means that only sensor raw domain basic noise reduction is enabled ,to remove demosaicing or other processing artifacts. For YUV_REPROCESSING, MINIMAL is same as OFF. This mode is optional, may not be support by all devices. The application should check android.noiseReduction.availableNoiseReductionModes before using it.

FAST/HIGH_QUALITY both mean camera device determined noise filtering will be applied. HIGH_QUALITY mode indicates that the camera device will use the highest-quality noise filtering algorithms, even if it slows down capture rate. FAST means the camera device will not slow down capture rate when applying noise filtering. FAST may be the same as MINIMAL if MINIMAL is listed, or the same as OFF if any noise filtering will slow down capture rate. Every output stream will have a similar amount of enhancement applied.

ZERO_SHUTTER_LAG is meant to be used by applications that maintain a continuous circular buffer of high-resolution images during preview and reprocess image(s) from that buffer into a final capture when triggered by the user. In this mode, the camera device applies noise reduction to low-resolution streams (below maximum recording resolution) to maximize preview quality, but does not apply noise reduction to high-resolution streams, since those will be reprocessed later if necessary.

For YUV_REPROCESSING, these FAST/HIGH_QUALITY modes both mean that the camera device will apply FAST/HIGH_QUALITY YUV domain noise reduction, respectively. The camera device may adjust the noise reduction parameters for best image quality based on the android.reprocess.effectiveExposureFactor if it is set.

HAL Implementation Details

For YUV_REPROCESSING The HAL can use android.reprocess.effectiveExposureFactor to adjust the internal noise reduction parameters appropriately to get the best quality images.

quirks
static
Property Name Type Description Units Range Tags
android.quirks.meteringCropRegion byte [system] [deprecated]

If set to 1, the camera service does not scale 'normalized' coordinates with respect to the crop region. This applies to metering input (a{e,f,wb}Region and output (face rectangles).

Deprecated. Do not use.

Details

Normalized coordinates refer to those in the (-1000,1000) range mentioned in the android.hardware.Camera API.

HAL implementations should instead always use and emit sensor array-relative coordinates for all region data. Does not need to be listed in static metadata. Support will be removed in future versions of camera service.

android.quirks.triggerAfWithAuto byte [system] [deprecated]

If set to 1, then the camera service always switches to FOCUS_MODE_AUTO before issuing a AF trigger.

Deprecated. Do not use.

Details

HAL implementations should implement AF trigger modes for AUTO, MACRO, CONTINUOUS_FOCUS, and CONTINUOUS_PICTURE modes instead of using this flag. Does not need to be listed in static metadata. Support will be removed in future versions of camera service

android.quirks.useZslFormat byte [system] [deprecated]

If set to 1, the camera service uses CAMERA2_PIXEL_FORMAT_ZSL instead of HAL_PIXEL_FORMAT_IMPLEMENTATION_DEFINED for the zero shutter lag stream

Deprecated. Do not use.

Details

HAL implementations should use gralloc usage flags to determine that a stream will be used for zero-shutter-lag, instead of relying on an explicit format setting. Does not need to be listed in static metadata. Support will be removed in future versions of camera service.

android.quirks.usePartialResult byte [hidden] [deprecated]

If set to 1, the HAL will always split result metadata for a single capture into multiple buffers, returned using multiple process_capture_result calls.

Deprecated. Do not use.

Details

Does not need to be listed in static metadata. Support for partial results will be reworked in future versions of camera service. This quirk will stop working at that point; DO NOT USE without careful consideration of future support.

HAL Implementation Details

Refer to camera3_capture_result::partial_result for information on how to implement partial results.

dynamic
Property Name Type Description Units Range Tags
android.quirks.partialResult byte [hidden as boolean] [deprecated]
  • FINAL

    The last or only metadata result buffer for this capture.

  • PARTIAL

    A partial buffer of result metadata for this capture. More result buffers for this capture will be sent by the camera device, the last of which will be marked FINAL.

Whether a result given to the framework is the final one for the capture, or only a partial that contains a subset of the full set of dynamic metadata values.

Deprecated. Do not use.

Optional. Default value is FINAL.

Details

The entries in the result metadata buffers for a single capture may not overlap, except for this entry. The FINAL buffers must retain FIFO ordering relative to the requests that generate them, so the FINAL buffer for frame 3 must always be sent to the framework after the FINAL buffer for frame 2, and before the FINAL buffer for frame 4. PARTIAL buffers may be returned in any order relative to other frames, but all PARTIAL buffers for a given capture must arrive before the FINAL buffer for that capture. This entry may only be used by the camera device if quirks.usePartialResult is set to 1.

HAL Implementation Details

Refer to camera3_capture_result::partial_result for information on how to implement partial results.

request
controls
Property Name Type Description Units Range Tags
android.request.frameCount int32 [system] [deprecated]

A frame counter set by the framework. Must be maintained unchanged in output frame. This value monotonically increases with every new result (that is, each new result has a unique frameCount value).

incrementing integer

Deprecated. Do not use.

Any int.

android.request.id int32 [hidden]

An application-specified ID for the current request. Must be maintained unchanged in output frame

arbitrary integer assigned by application

Any int

android.request.inputStreams int32 x n [system] [deprecated]

List which camera reprocess stream is used for the source of reprocessing data.

List of camera reprocess stream IDs

Deprecated. Do not use.

Typically, only one entry allowed, must be a valid reprocess stream ID.

Details

Only meaningful when android.request.type == REPROCESS. Ignored otherwise

android.request.metadataMode byte [system]
  • NONE

    No metadata should be produced on output, except for application-bound buffer data. If no application-bound streams exist, no frame should be placed in the output frame queue. If such streams exist, a frame should be placed on the output queue with null metadata but with the necessary output buffer information. Timestamp information should still be included with any output stream buffers

  • FULL

    All metadata should be produced. Statistics will only be produced if they are separately enabled

How much metadata to produce on output

android.request.outputStreams int32 x n [system] [deprecated]

Lists which camera output streams image data from this capture must be sent to

List of camera stream IDs

Deprecated. Do not use.

List must only include streams that have been created

Details

If no output streams are listed, then the image data should simply be discarded. The image data must still be captured for metadata and statistics production, and the lens and flash must operate as requested.

android.request.type byte [system] [deprecated]
  • CAPTURE

    Capture a new image from the imaging hardware, and process it according to the settings

  • REPROCESS

    Process previously captured data; the android.request.inputStreams parameter determines the source reprocessing stream. TODO: Mark dynamic metadata needed for reprocessing with [RP]

The type of the request; either CAPTURE or REPROCESS. For HAL3, this tag is redundant.

Deprecated. Do not use.

static
Property Name Type Description Units Range Tags
android.request.maxNumOutputStreams int32 x 3 [hidden] [legacy]

The maximum numbers of different types of output streams that can be configured and used simultaneously by a camera device.

For processed (and stalling) format streams, >= 1.

For Raw format (either stalling or non-stalling) streams, >= 0.

For processed (but not stalling) format streams, >= 3 for FULL mode devices (android.info.supportedHardwareLevel == FULL); >= 2 for LIMITED mode devices (android.info.supportedHardwareLevel == LIMITED).

Details

This is a 3 element tuple that contains the max number of output simultaneous streams for raw sensor, processed (but not stalling), and processed (and stalling) formats respectively. For example, assuming that JPEG is typically a processed and stalling stream, if max raw sensor format output stream number is 1, max YUV streams number is 3, and max JPEG stream number is 2, then this tuple should be (1, 3, 2).

This lists the upper bound of the number of output streams supported by the camera device. Using more streams simultaneously may require more hardware and CPU resources that will consume more power. The image format for an output stream can be any supported format provided by android.scaler.availableStreamConfigurations. The formats defined in android.scaler.availableStreamConfigurations can be catergorized into the 3 stream types as below:

  • Processed (but stalling): any non-RAW format with a stallDurations > 0. Typically JPEG format.
  • Raw formats: RAW_SENSOR, RAW10, or RAW12.
  • Processed (but not-stalling): any non-RAW format without a stall duration. Typically YUV_420_888, NV21, or YV12.
android.request.maxNumOutputRaw int32 [public] [synthetic] [legacy]

The maximum numbers of different types of output streams that can be configured and used simultaneously by a camera device for any RAW formats.

>= 0

Details

This value contains the max number of output simultaneous streams from the raw sensor.

This lists the upper bound of the number of output streams supported by the camera device. Using more streams simultaneously may require more hardware and CPU resources that will consume more power. The image format for this kind of an output stream can be any RAW and supported format provided by android.scaler.streamConfigurationMap.

In particular, a RAW format is typically one of:

LEGACY mode devices (android.info.supportedHardwareLevel == LEGACY) never support raw streams.

android.request.maxNumOutputProc int32 [public] [synthetic] [legacy]

The maximum numbers of different types of output streams that can be configured and used simultaneously by a camera device for any processed (but not-stalling) formats.

>= 3 for FULL mode devices (android.info.supportedHardwareLevel == FULL); >= 2 for LIMITED mode devices (android.info.supportedHardwareLevel == LIMITED).

Details

This value contains the max number of output simultaneous streams for any processed (but not-stalling) formats.

This lists the upper bound of the number of output streams supported by the camera device. Using more streams simultaneously may require more hardware and CPU resources that will consume more power. The image format for this kind of an output stream can be any non-RAW and supported format provided by android.scaler.streamConfigurationMap.

Processed (but not-stalling) is defined as any non-RAW format without a stall duration. Typically:

For full guarantees, query StreamConfigurationMap#getOutputStallDuration with a processed format -- it will return 0 for a non-stalling stream.

LEGACY devices will support at least 2 processing/non-stalling streams.

android.request.maxNumOutputProcStalling int32 [public] [synthetic] [legacy]

The maximum numbers of different types of output streams that can be configured and used simultaneously by a camera device for any processed (and stalling) formats.

>= 1

Details

This value contains the max number of output simultaneous streams for any processed (but not-stalling) formats.

This lists the upper bound of the number of output streams supported by the camera device. Using more streams simultaneously may require more hardware and CPU resources that will consume more power. The image format for this kind of an output stream can be any non-RAW and supported format provided by android.scaler.streamConfigurationMap.

A processed and stalling format is defined as any non-RAW format with a stallDurations > 0. Typically only the JPEG format is a stalling format.

For full guarantees, query StreamConfigurationMap#getOutputStallDuration with a processed format -- it will return a non-0 value for a stalling stream.

LEGACY devices will support up to 1 processing/stalling stream.

android.request.maxNumReprocessStreams int32 x 1 [system] [deprecated]

How many reprocessing streams of any type can be allocated at the same time.

Deprecated. Do not use.

>= 0

Details

Only used by HAL2.x.

When set to 0, it means no reprocess stream is supported.

android.request.maxNumInputStreams int32 [public] [full]

The maximum numbers of any type of input streams that can be configured and used simultaneously by a camera device.

0 or 1.

Details

When set to 0, it means no input stream is supported.

The image format for a input stream can be any supported format returned by StreamConfigurationMap#getInputFormats. When using an input stream, there must be at least one output stream configured to to receive the reprocessed images.

When an input stream and some output streams are used in a reprocessing request, only the input buffer will be used to produce these output stream buffers, and a new sensor image will not be captured.

For example, for Zero Shutter Lag (ZSL) still capture use case, the input stream image format will be PRIVATE, the associated output stream image format should be JPEG.

HAL Implementation Details

For the reprocessing flow and controls, see hardware/libhardware/include/hardware/camera3.h Section 10 for more details.

android.request.pipelineMaxDepth byte [public] [legacy]

Specifies the number of maximum pipeline stages a frame has to go through from when it's exposed to when it's available to the framework.

Details

A typical minimum value for this is 2 (one stage to expose, one stage to readout) from the sensor. The ISP then usually adds its own stages to do custom HW processing. Further stages may be added by SW processing.

Depending on what settings are used (e.g. YUV, JPEG) and what processing is enabled (e.g. face detection), the actual pipeline depth (specified by android.request.pipelineDepth) may be less than the max pipeline depth.

A pipeline depth of X stages is equivalent to a pipeline latency of X frame intervals.

This value will normally be 8 or less, however, for high speed capture session, the max pipeline depth will be up to 8 x size of high speed capture request list.

HAL Implementation Details

This value should be 4 or less, expect for the high speed recording session, where the max batch sizes may be larger than 1.

android.request.partialResultCount int32 [public]

Defines how many sub-components a result will be composed of.

>= 1

Details

In order to combat the pipeline latency, partial results may be delivered to the application layer from the camera device as soon as they are available.

Optional; defaults to 1. A value of 1 means that partial results are not supported, and only the final TotalCaptureResult will be produced by the camera device.

A typical use case for this might be: after requesting an auto-focus (AF) lock the new AF state might be available 50% of the way through the pipeline. The camera device could then immediately dispatch this state via a partial result to the application, and the rest of the metadata via later partial results.

android.request.availableCapabilities byte x n [public] [legacy]

List of capabilities that this camera device advertises as fully supporting.

Details

A capability is a contract that the camera device makes in order to be able to satisfy one or more use cases.

Listing a capability guarantees that the whole set of features required to support a common use will all be available.

Using a subset of the functionality provided by an unsupported capability may be possible on a specific camera device implementation; to do this query each of android.request.availableRequestKeys, android.request.availableResultKeys, android.request.availableCharacteristicsKeys.

The following capabilities are guaranteed to be available on android.info.supportedHardwareLevel == FULL devices:

  • MANUAL_SENSOR
  • MANUAL_POST_PROCESSING

Other capabilities may be available on either FULL or LIMITED devices, but the application should query this key to be sure.

HAL Implementation Details

Additional constraint details per-capability will be available in the Compatibility Test Suite.

Minimum baseline requirements required for the BACKWARD_COMPATIBLE capability are not explicitly listed. Instead refer to "BC" tags and the camera CTS tests in the android.hardware.camera2.cts package.

Listed controls that can be either request or result (e.g. android.sensor.exposureTime) must be available both in the request and the result in order to be considered to be capability-compliant.

For example, if the HAL claims to support MANUAL control, then exposure time must be configurable via the request and the actual exposure applied must be available via the result.

If MANUAL_SENSOR is omitted, the HAL may choose to omit the android.scaler.availableMinFrameDurations static property entirely.

For PRIVATE_REPROCESSING and YUV_REPROCESSING capabilities, see hardware/libhardware/include/hardware/camera3.h Section 10 for more information.

Devices that support the MANUAL_SENSOR capability must support the CAMERA3_TEMPLATE_MANUAL template defined in camera3.h.

Devices that support the PRIVATE_REPROCESSING capability or the YUV_REPROCESSING capability must support the CAMERA3_TEMPLATE_ZERO_SHUTTER_LAG template defined in camera3.h.

For DEPTH_OUTPUT, the depth-format keys android.depth.availableDepthStreamConfigurations, android.depth.availableDepthMinFrameDurations, android.depth.availableDepthStallDurations must be available, in addition to the other keys explicitly mentioned in the DEPTH_OUTPUT enum notes. The entry android.depth.maxDepthSamples must be available if the DEPTH_POINT_CLOUD format is supported (HAL pixel format BLOB, dataspace DEPTH).

android.request.availableRequestKeys int32 x n [hidden] [legacy]

A list of all keys that the camera device has available to use with CaptureRequest.

Details

Attempting to set a key into a CaptureRequest that is not listed here will result in an invalid request and will be rejected by the camera device.

This field can be used to query the feature set of a camera device at a more granular level than capabilities. This is especially important for optional keys that are not listed under any capability in android.request.availableCapabilities.

HAL Implementation Details

Vendor tags must not be listed here. Use the vendor tag metadata extensions C api instead (refer to camera3.h for more details).

Setting/getting vendor tags will be checked against the metadata vendor extensions API and not against this field.

The HAL must not consume any request tags that are not listed either here or in the vendor tag list.

The public camera2 API will always make the vendor tags visible via CameraCharacteristics#getAvailableCaptureRequestKeys.

android.request.availableResultKeys int32 x n [hidden] [legacy]

A list of all keys that the camera device has available to use with CaptureResult.

Details

Attempting to get a key from a CaptureResult that is not listed here will always return a null value. Getting a key from a CaptureResult that is listed here will generally never return a null value.

The following keys may return null unless they are enabled:

(Those sometimes-null keys will nevertheless be listed here if they are available.)

This field can be used to query the feature set of a camera device at a more granular level than capabilities. This is especially important for optional keys that are not listed under any capability in android.request.availableCapabilities.

HAL Implementation Details

Tags listed here must always have an entry in the result metadata, even if that size is 0 elements. Only array-type tags (e.g. lists, matrices, strings) are allowed to have 0 elements.

Vendor tags must not be listed here. Use the vendor tag metadata extensions C api instead (refer to camera3.h for more details).

Setting/getting vendor tags will be checked against the metadata vendor extensions API and not against this field.

The HAL must not produce any result tags that are not listed either here or in the vendor tag list.

The public camera2 API will always make the vendor tags visible via CameraCharacteristics#getAvailableCaptureResultKeys.

android.request.availableCharacteristicsKeys int32 x n [hidden] [legacy]

A list of all keys that the camera device has available to use with CameraCharacteristics.

Details

This entry follows the same rules as android.request.availableResultKeys (except that it applies for CameraCharacteristics instead of CaptureResult). See above for more details.

HAL Implementation Details

Keys listed here must always have an entry in the static info metadata, even if that size is 0 elements. Only array-type tags (e.g. lists, matrices, strings) are allowed to have 0 elements.

Vendor tags must not be listed here. Use the vendor tag metadata extensions C api instead (refer to camera3.h for more details).

Setting/getting vendor tags will be checked against the metadata vendor extensions API and not against this field.

The HAL must not have any tags in its static info that are not listed either here or in the vendor tag list.

The public camera2 API will always make the vendor tags visible via CameraCharacteristics#getKeys.

dynamic
Property Name Type Description Units Range Tags
android.request.frameCount int32 [hidden] [deprecated]

A frame counter set by the framework. This value monotonically increases with every new result (that is, each new result has a unique frameCount value).

count of frames

Deprecated. Do not use.

> 0

Details

Reset on release()

android.request.id int32 [hidden]

An application-specified ID for the current request. Must be maintained unchanged in output frame

arbitrary integer assigned by application

Any int

android.request.metadataMode byte [system]
  • NONE

    No metadata should be produced on output, except for application-bound buffer data. If no application-bound streams exist, no frame should be placed in the output frame queue. If such streams exist, a frame should be placed on the output queue with null metadata but with the necessary output buffer information. Timestamp information should still be included with any output stream buffers

  • FULL

    All metadata should be produced. Statistics will only be produced if they are separately enabled

How much metadata to produce on output

android.request.outputStreams int32 x n [system] [deprecated]

Lists which camera output streams image data from this capture must be sent to

List of camera stream IDs

Deprecated. Do not use.

List must only include streams that have been created

Details

If no output streams are listed, then the image data should simply be discarded. The image data must still be captured for metadata and statistics production, and the lens and flash must operate as requested.

android.request.pipelineDepth byte [public] [legacy]

Specifies the number of pipeline stages the frame went through from when it was exposed to when the final completed result was available to the framework.

<= android.request.pipelineMaxDepth

Details

Depending on what settings are used in the request, and what streams are configured, the data may undergo less processing, and some pipeline stages skipped.

See android.request.pipelineMaxDepth for more details.

HAL Implementation Details

This value must always represent the accurate count of how many pipeline stages were actually used.

scaler
controls
Property Name Type Description Units Range Tags
android.scaler.cropRegion int32 x 4 [public as rectangle] [legacy]

The desired region of the sensor to read out for this capture.

Pixel coordinates relative to android.sensor.info.activeArraySize
Details

This control can be used to implement digital zoom.

The crop region coordinate system is based off android.sensor.info.activeArraySize, with (0, 0) being the top-left corner of the sensor active array.

Output streams use this rectangle to produce their output, cropping to a smaller region if necessary to maintain the stream's aspect ratio, then scaling the sensor input to match the output's configured resolution.

The crop region is applied after the RAW to other color space (e.g. YUV) conversion. Since raw streams (e.g. RAW16) don't have the conversion stage, they are not croppable. The crop region will be ignored by raw streams.

For non-raw streams, any additional per-stream cropping will be done to maximize the final pixel area of the stream.

For example, if the crop region is set to a 4:3 aspect ratio, then 4:3 streams will use the exact crop region. 16:9 streams will further crop vertically (letterbox).

Conversely, if the crop region is set to a 16:9, then 4:3 outputs will crop horizontally (pillarbox), and 16:9 streams will match exactly. These additional crops will be centered within the crop region.

The width and height of the crop region cannot be set to be smaller than floor( activeArraySize.width / android.scaler.availableMaxDigitalZoom ) and floor( activeArraySize.height / android.scaler.availableMaxDigitalZoom ), respectively.

The camera device may adjust the crop region to account for rounding and other hardware requirements; the final crop region used will be included in the output capture result.

HAL Implementation Details

The output streams must maintain square pixels at all times, no matter what the relative aspect ratios of the crop region and the stream are. Negative values for corner are allowed for raw output if full pixel array is larger than active pixel array. Width and height may be rounded to nearest larger supportable width, especially for raw output, where only a few fixed scales may be possible.

For a set of output streams configured, if the sensor output is cropped to a smaller size than active array size, the HAL need follow below cropping rules:

  • The HAL need handle the cropRegion as if the sensor crop size is the effective active array size.More specifically, the HAL must transform the request cropRegion from android.sensor.info.activeArraySize to the sensor cropped pixel area size in this way:

    1. Translate the requested cropRegion w.r.t., the left top corner of the sensor cropped pixel area by (tx, ty), where tx = sensorCrop.top * (sensorCrop.height / activeArraySize.height) and tx = sensorCrop.left * (sensorCrop.width / activeArraySize.width). The (sensorCrop.top, sensorCrop.left) is the coordinate based off the android.sensor.info.activeArraySize.
    2. Scale the width and height of requested cropRegion with scaling factor of sensorCrop.width/activeArraySize.width and sensorCrop.height/activeArraySize.height respectively. Once this new cropRegion is calculated, the HAL must use this region to crop the image with regard to the sensor crop size (effective active array size). The HAL still need follow the general cropping rule for this new cropRegion and effective active array size.
  • The HAL must report the cropRegion with regard to android.sensor.info.activeArraySize. The HAL need convert the new cropRegion generated above w.r.t., full active array size. The reported cropRegion may be slightly different with the requested cropRegion since the HAL may adjust the crop region to account for rounding, conversion error, or other hardware limitations.

HAL2.x uses only (x, y, width)

static
Property Name Type Description Units Range Tags
android.scaler.availableFormats int32 x n [hidden as imageFormat] [deprecated]
  • RAW16 [optional] 0x20

    RAW16 is a standard, cross-platform format for raw image buffers with 16-bit pixels.

    Buffers of this format are typically expected to have a Bayer Color Filter Array (CFA) layout, which is given in android.sensor.info.colorFilterArrangement. Sensors with CFAs that are not representable by a format in android.sensor.info.colorFilterArrangement should not use this format.

    Buffers of this format will also follow the constraints given for RAW_OPAQUE buffers, but with relaxed performance constraints.

    This format is intended to give users access to the full contents of the buffers coming directly from the image sensor prior to any cropping or scaling operations, and all coordinate systems for metadata used for this format are relative to the size of the active region of the image sensor before any geometric distortion correction has been applied (i.e. android.sensor.info.preCorrectionActiveArraySize). Supported dimensions for this format are limited to the full dimensions of the sensor (e.g. either android.sensor.info.pixelArraySize or android.sensor.info.preCorrectionActiveArraySize will be the only supported output size).

    See android.scaler.availableInputOutputFormatsMap for the full set of performance guarantees.

  • RAW_OPAQUE [optional] 0x24

    RAW_OPAQUE is a format for raw image buffers coming from an image sensor.

    The actual structure of buffers of this format is platform-specific, but must follow several constraints:

    1. No image post-processing operations may have been applied to buffers of this type. These buffers contain raw image data coming directly from the image sensor.
    2. If a buffer of this format is passed to the camera device for reprocessing, the resulting images will be identical to the images produced if the buffer had come directly from the sensor and was processed with the same settings.

    The intended use for this format is to allow access to the native raw format buffers coming directly from the camera sensor without any additional conversions or decrease in framerate.

    See android.scaler.availableInputOutputFormatsMap for the full set of performance guarantees.

  • YV12 [optional] 0x32315659

    YCrCb 4:2:0 Planar

  • YCrCb_420_SP [optional] 0x11

    NV21

  • IMPLEMENTATION_DEFINED 0x22

    System internal format, not application-accessible

  • YCbCr_420_888 0x23

    Flexible YUV420 Format

  • BLOB 0x21

    JPEG format

The list of image formats that are supported by this camera device for output streams.

Deprecated. Do not use.

Details

All camera devices will support JPEG and YUV_420_888 formats.

When set to YUV_420_888, application can access the YUV420 data directly.

HAL Implementation Details

These format values are from HAL_PIXEL_FORMAT_* in system/core/include/system/graphics.h.

When IMPLEMENTATION_DEFINED is used, the platform gralloc module will select a format based on the usage flags provided by the camera HAL device and the other endpoint of the stream. It is usually used by preview and recording streams, where the application doesn't need access the image data.

YCbCr_420_888 format must be supported by the HAL. When an image stream needs CPU/application direct access, this format will be used.

The BLOB format must be supported by the HAL. This is used for the JPEG stream.

A RAW_OPAQUE buffer should contain only pixel data. It is strongly recommended that any information used by the camera device when processing images is fully expressed by the result metadata for that image buffer.

android.scaler.availableJpegMinDurations int64 x n [hidden] [deprecated]

The minimum frame duration that is supported for each resolution in android.scaler.availableJpegSizes.

Nanoseconds

Deprecated. Do not use.

TODO: Remove property.

Details

This corresponds to the minimum steady-state frame duration when only that JPEG stream is active and captured in a burst, with all processing (typically in android.*.mode) set to FAST.

When multiple streams are configured, the minimum frame duration will be >= max(individual stream min durations)

android.scaler.availableJpegSizes int32 x n x 2 [hidden as size] [deprecated]

The JPEG resolutions that are supported by this camera device.

Deprecated. Do not use.

TODO: Remove property.

Details

The resolutions are listed as (width, height) pairs. All camera devices will support sensor maximum resolution (defined by android.sensor.info.activeArraySize).

HAL Implementation Details

The HAL must include sensor maximum resolution (defined by android.sensor.info.activeArraySize), and should include half/quarter of sensor maximum resolution.

android.scaler.availableMaxDigitalZoom float [public] [legacy]

The maximum ratio between both active area width and crop region width, and active area height and crop region height, for android.scaler.cropRegion.

Zoom scale factor

>=1

Details

This represents the maximum amount of zooming possible by the camera device, or equivalently, the minimum cropping window size.

Crop regions that have a width or height that is smaller than this ratio allows will be rounded up to the minimum allowed size by the camera device.

android.scaler.availableProcessedMinDurations int64 x n [hidden] [deprecated]

For each available processed output size (defined in android.scaler.availableProcessedSizes), this property lists the minimum supportable frame duration for that size.

Nanoseconds

Deprecated. Do not use.

Details

This should correspond to the frame duration when only that processed stream is active, with all processing (typically in android.*.mode) set to FAST.

When multiple streams are configured, the minimum frame duration will be >= max(individual stream min durations).

android.scaler.availableProcessedSizes int32 x n x 2 [hidden as size] [deprecated]

The resolutions available for use with processed output streams, such as YV12, NV12, and platform opaque YUV/RGB streams to the GPU or video encoders.

Deprecated. Do not use.

Details

The resolutions are listed as (width, height) pairs.

For a given use case, the actual maximum supported resolution may be lower than what is listed here, depending on the destination Surface for the image data. For example, for recording video, the video encoder chosen may have a maximum size limit (e.g. 1080p) smaller than what the camera (e.g. maximum resolution is 3264x2448) can provide.

Please reference the documentation for the image data destination to check if it limits the maximum size for image data.

HAL Implementation Details

For FULL capability devices (android.info.supportedHardwareLevel == FULL), the HAL must include all JPEG sizes listed in android.scaler.availableJpegSizes and each below resolution if it is smaller than or equal to the sensor maximum resolution (if they are not listed in JPEG sizes already):

  • 240p (320 x 240)
  • 480p (640 x 480)
  • 720p (1280 x 720)
  • 1080p (1920 x 1080)

For LIMITED capability devices (android.info.supportedHardwareLevel == LIMITED), the HAL only has to list up to the maximum video size supported by the devices.

android.scaler.availableRawMinDurations int64 x n [system] [deprecated]

For each available raw output size (defined in android.scaler.availableRawSizes), this property lists the minimum supportable frame duration for that size.

Nanoseconds

Deprecated. Do not use.

Details

Should correspond to the frame duration when only the raw stream is active.

When multiple streams are configured, the minimum frame duration will be >= max(individual stream min durations)

android.scaler.availableRawSizes int32 x n x 2 [system as size] [deprecated]

The resolutions available for use with raw sensor output streams, listed as width, height

Deprecated. Do not use.

android.scaler.availableInputOutputFormatsMap int32 [hidden as reprocessFormatsMap]

The mapping of image formats that are supported by this camera device for input streams, to their corresponding output formats.

Details

All camera devices with at least 1 android.request.maxNumInputStreams will have at least one available input format.

The camera device will support the following map of formats, if its dependent capability (android.request.availableCapabilities) is supported:

Input Format Output Format Capability
ImageFormat#PRIVATE ImageFormat#JPEG PRIVATE_REPROCESSING
ImageFormat#PRIVATE ImageFormat#YUV_420_888 PRIVATE_REPROCESSING
ImageFormat#YUV_420_888 ImageFormat#JPEG YUV_REPROCESSING
ImageFormat#YUV_420_888 ImageFormat#YUV_420_888 YUV_REPROCESSING

PRIVATE refers to a device-internal format that is not directly application-visible. A PRIVATE input surface can be acquired by ImageReader#newInstance with ImageFormat#PRIVATE as the format.

For a PRIVATE_REPROCESSING-capable camera device, using the PRIVATE format as either input or output will never hurt maximum frame rate (i.e. getOutputStallDuration(ImageFormat.PRIVATE, size) is always 0),

Attempting to configure an input stream with output streams not listed as available in this map is not valid.

HAL Implementation Details

For the formats, see system/core/include/system/graphics.h for a definition of the image format enumerations. The PRIVATE format refers to the HAL_PIXEL_FORMAT_IMPLEMENTATION_DEFINED format. The HAL could determine the actual format by using the gralloc usage flags. For ZSL use case in particular, the HAL could choose appropriate format (partially processed YUV or RAW based format) by checking the format and GRALLOC_USAGE_HW_CAMERA_ZSL. See camera3.h for more details.

This value is encoded as a variable-size array-of-arrays. The inner array always contains [format, length, ...] where ... has length elements. An inner array is followed by another inner array if the total metadata entry size hasn't yet been exceeded.

A code sample to read/write this encoding (with a device that supports reprocessing IMPLEMENTATION_DEFINED to YUV_420_888, and JPEG, and reprocessing YUV_420_888 to YUV_420_888 and JPEG):

// reading
int32_t* contents = &entry.i32[0];
for (size_t i = 0; i < entry.count; ) {
    int32_t format = contents[i++];
    int32_t length = contents[i++];
    int32_t output_formats[length];
    memcpy(&output_formats[0], &contents[i],
           length * sizeof(int32_t));
    i += length;
}

// writing (static example, PRIVATE_REPROCESSING + YUV_REPROCESSING)
int32_t[] contents = {
  IMPLEMENTATION_DEFINED, 2, YUV_420_888, BLOB,
  YUV_420_888, 2, YUV_420_888, BLOB,
};
update_camera_metadata_entry(metadata, index, &contents[0],
      sizeof(contents)/sizeof(contents[0]), &updated_entry);

If the HAL claims to support any of the capabilities listed in the above details, then it must also support all the input-output combinations listed for that capability. It can optionally support additional formats if it so chooses.

android.scaler.availableStreamConfigurations int32 x n x 4 [hidden as streamConfiguration] [legacy]
  • OUTPUT
  • INPUT

The available stream configurations that this camera device supports (i.e. format, width, height, output/input stream).

Details

The configurations are listed as (format, width, height, input?) tuples.

For a given use case, the actual maximum supported resolution may be lower than what is listed here, depending on the destination Surface for the image data. For example, for recording video, the video encoder chosen may have a maximum size limit (e.g. 1080p) smaller than what the camera (e.g. maximum resolution is 3264x2448) can provide.

Please reference the documentation for the image data destination to check if it limits the maximum size for image data.

Not all output formats may be supported in a configuration with an input stream of a particular format. For more details, see android.scaler.availableInputOutputFormatsMap.

The following table describes the minimum required output stream configurations based on the hardware level (android.info.supportedHardwareLevel):

Format Size Hardware Level Notes
JPEG android.sensor.info.activeArraySize Any
JPEG 1920x1080 (1080p) Any if 1080p <= activeArraySize
JPEG 1280x720 (720) Any if 720p <= activeArraySize
JPEG 640x480 (480p) Any if 480p <= activeArraySize
JPEG 320x240 (240p) Any if 240p <= activeArraySize
YUV_420_888 all output sizes available for JPEG FULL
YUV_420_888 all output sizes available for JPEG, up to the maximum video size LIMITED
IMPLEMENTATION_DEFINED same as YUV_420_888 Any

Refer to android.request.availableCapabilities for additional mandatory stream configurations on a per-capability basis.

HAL Implementation Details

It is recommended (but not mandatory) to also include half/quarter of sensor maximum resolution for JPEG formats (regardless of hardware level).

(The following is a rewording of the above required table):

For JPEG format, the sizes may be restricted by below conditions:

  • The HAL may choose the aspect ratio of each Jpeg size to be one of well known ones (e.g. 4:3, 16:9, 3:2 etc.). If the sensor maximum resolution (defined by android.sensor.info.activeArraySize) has an aspect ratio other than these, it does not have to be included in the supported JPEG sizes.
  • Some hardware JPEG encoders may have pixel boundary alignment requirements, such as the dimensions being a multiple of 16.

Therefore, the maximum JPEG size may be smaller than sensor maximum resolution. However, the largest JPEG size must be as close as possible to the sensor maximum resolution given above constraints. It is required that after aspect ratio adjustments, additional size reduction due to other issues must be less than 3% in area. For example, if the sensor maximum resolution is 3280x2464, if the maximum JPEG size has aspect ratio 4:3, the JPEG encoder alignment requirement is 16, the maximum JPEG size will be 3264x2448.

For FULL capability devices (android.info.supportedHardwareLevel == FULL), the HAL must include all YUV_420_888 sizes that have JPEG sizes listed here as output streams.

It must also include each below resolution if it is smaller than or equal to the sensor maximum resolution (for both YUV_420_888 and JPEG formats), as output streams:

  • 240p (320 x 240)
  • 480p (640 x 480)
  • 720p (1280 x 720)
  • 1080p (1920 x 1080)

For LIMITED capability devices (android.info.supportedHardwareLevel == LIMITED), the HAL only has to list up to the maximum video size supported by the device.

Regardless of hardware level, every output resolution available for YUV_420_888 must also be available for IMPLEMENTATION_DEFINED.

This supercedes the following fields, which are now deprecated:

  • availableFormats
  • available[Processed,Raw,Jpeg]Sizes
android.scaler.availableMinFrameDurations int64 x 4 x n [hidden as streamConfigurationDuration] [legacy]

This lists the minimum frame duration for each format/size combination.

(format, width, height, ns) x n
Details

This should correspond to the frame duration when only that stream is active, with all processing (typically in android.*.mode) set to either OFF or FAST.

When multiple streams are used in a request, the minimum frame duration will be max(individual stream min durations).

The minimum frame duration of a stream (of a particular format, size) is the same regardless of whether the stream is input or output.

See android.sensor.frameDuration and android.scaler.availableStallDurations for more details about calculating the max frame rate.

(Keep in sync with StreamConfigurationMap#getOutputMinFrameDuration)

android.scaler.availableStallDurations int64 x 4 x n [hidden as streamConfigurationDuration] [legacy]

This lists the maximum stall duration for each output format/size combination.

(format, width, height, ns) x n
Details

A stall duration is how much extra time would get added to the normal minimum frame duration for a repeating request that has streams with non-zero stall.

For example, consider JPEG captures which have the following characteristics:

  • JPEG streams act like processed YUV streams in requests for which they are not included; in requests in which they are directly referenced, they act as JPEG streams. This is because supporting a JPEG stream requires the underlying YUV data to always be ready for use by a JPEG encoder, but the encoder will only be used (and impact frame duration) on requests that actually reference a JPEG stream.
  • The JPEG processor can run concurrently to the rest of the camera pipeline, but cannot process more than 1 capture at a time.

In other words, using a repeating YUV request would result in a steady frame rate (let's say it's 30 FPS). If a single JPEG request is submitted periodically, the frame rate will stay at 30 FPS (as long as we wait for the previous JPEG to return each time). If we try to submit a repeating YUV + JPEG request, then the frame rate will drop from 30 FPS.

In general, submitting a new request with a non-0 stall time stream will not cause a frame rate drop unless there are still outstanding buffers for that stream from previous requests.

Submitting a repeating request with streams (call this S) is the same as setting the minimum frame duration from the normal minimum frame duration corresponding to S, added with the maximum stall duration for S.

If interleaving requests with and without a stall duration, a request will stall by the maximum of the remaining times for each can-stall stream with outstanding buffers.

This means that a stalling request will not have an exposure start until the stall has completed.

This should correspond to the stall duration when only that stream is active, with all processing (typically in android.*.mode) set to FAST or OFF. Setting any of the processing modes to HIGH_QUALITY effectively results in an indeterminate stall duration for all streams in a request (the regular stall calculation rules are ignored).

The following formats may always have a stall duration:

The following formats will never have a stall duration:

All other formats may or may not have an allowed stall duration on a per-capability basis; refer to android.request.availableCapabilities for more details.

See android.sensor.frameDuration for more information about calculating the max frame rate (absent stalls).

(Keep up to date with StreamConfigurationMap#getOutputStallDuration )

HAL Implementation Details

If possible, it is recommended that all non-JPEG formats (such as RAW16) should not have a stall duration. RAW10, RAW12, RAW_OPAQUE and IMPLEMENTATION_DEFINED must not have stall durations.

android.scaler.streamConfigurationMap int32 [public as streamConfigurationMap] [synthetic] [legacy]

The available stream configurations that this camera device supports; also includes the minimum frame durations and the stall durations for each format/size combination.

Details

All camera devices will support sensor maximum resolution (defined by android.sensor.info.activeArraySize) for the JPEG format.

For a given use case, the actual maximum supported resolution may be lower than what is listed here, depending on the destination Surface for the image data. For example, for recording video, the video encoder chosen may have a maximum size limit (e.g. 1080p) smaller than what the camera (e.g. maximum resolution is 3264x2448) can provide.

Please reference the documentation for the image data destination to check if it limits the maximum size for image data.

The following table describes the minimum required output stream configurations based on the hardware level (android.info.supportedHardwareLevel):

Format Size Hardware Level Notes
ImageFormat#JPEG android.sensor.info.activeArraySize (*1) Any
ImageFormat#JPEG 1920x1080 (1080p) Any if 1080p <= activeArraySize
ImageFormat#JPEG 1280x720 (720p) Any if 720p <= activeArraySize
ImageFormat#JPEG 640x480 (480p) Any if 480p <= activeArraySize
ImageFormat#JPEG 320x240 (240p) Any if 240p <= activeArraySize
ImageFormat#YUV_420_888 all output sizes available for JPEG FULL
ImageFormat#YUV_420_888 all output sizes available for JPEG, up to the maximum video size LIMITED
ImageFormat#PRIVATE same as YUV_420_888 Any

Refer to android.request.availableCapabilities and CameraDevice#createCaptureSession for additional mandatory stream configurations on a per-capability basis.

*1: For JPEG format, the sizes may be restricted by below conditions:

  • The HAL may choose the aspect ratio of each Jpeg size to be one of well known ones (e.g. 4:3, 16:9, 3:2 etc.). If the sensor maximum resolution (defined by android.sensor.info.activeArraySize) has an aspect ratio other than these, it does not have to be included in the supported JPEG sizes.
  • Some hardware JPEG encoders may have pixel boundary alignment requirements, such as the dimensions being a multiple of 16. Therefore, the maximum JPEG size may be smaller than sensor maximum resolution. However, the largest JPEG size will be as close as possible to the sensor maximum resolution given above constraints. It is required that after aspect ratio adjustments, additional size reduction due to other issues must be less than 3% in area. For example, if the sensor maximum resolution is 3280x2464, if the maximum JPEG size has aspect ratio 4:3, and the JPEG encoder alignment requirement is 16, the maximum JPEG size will be 3264x2448.
HAL Implementation Details

Do not set this property directly (it is synthetic and will not be available at the HAL layer); set the android.scaler.availableStreamConfigurations instead.

Not all output formats may be supported in a configuration with an input stream of a particular format. For more details, see android.scaler.availableInputOutputFormatsMap.

It is recommended (but not mandatory) to also include half/quarter of sensor maximum resolution for JPEG formats (regardless of hardware level).

(The following is a rewording of the above required table):

The HAL must include sensor maximum resolution (defined by android.sensor.info.activeArraySize).

For FULL capability devices (android.info.supportedHardwareLevel == FULL), the HAL must include all YUV_420_888 sizes that have JPEG sizes listed here as output streams.

It must also include each below resolution if it is smaller than or equal to the sensor maximum resolution (for both YUV_420_888 and JPEG formats), as output streams:

  • 240p (320 x 240)
  • 480p (640 x 480)
  • 720p (1280 x 720)
  • 1080p (1920 x 1080)

For LIMITED capability devices (android.info.supportedHardwareLevel == LIMITED), the HAL only has to list up to the maximum video size supported by the device.

Regardless of hardware level, every output resolution available for YUV_420_888 must also be available for IMPLEMENTATION_DEFINED.

This supercedes the following fields, which are now deprecated:

  • availableFormats
  • available[Processed,Raw,Jpeg]Sizes
android.scaler.croppingType byte [public] [legacy]
  • CENTER_ONLY

    The camera device only supports centered crop regions.

  • FREEFORM

    The camera device supports arbitrarily chosen crop regions.

The crop type that this camera device supports.

Details

When passing a non-centered crop region (android.scaler.cropRegion) to a camera device that only supports CENTER_ONLY cropping, the camera device will move the crop region to the center of the sensor active array (android.sensor.info.activeArraySize) and keep the crop region width and height unchanged. The camera device will return the final used crop region in metadata result android.scaler.cropRegion.

Camera devices that support FREEFORM cropping will support any crop region that is inside of the active array. The camera device will apply the same crop region and return the final used crop region in capture result metadata android.scaler.cropRegion.

LEGACY capability devices will only support CENTER_ONLY cropping.

dynamic
Property Name Type Description Units Range Tags
android.scaler.cropRegion int32 x 4 [public as rectangle] [legacy]

The desired region of the sensor to read out for this capture.

Pixel coordinates relative to android.sensor.info.activeArraySize
Details

This control can be used to implement digital zoom.

The crop region coordinate system is based off android.sensor.info.activeArraySize, with (0, 0) being the top-left corner of the sensor active array.

Output streams use this rectangle to produce their output, cropping to a smaller region if necessary to maintain the stream's aspect ratio, then scaling the sensor input to match the output's configured resolution.

The crop region is applied after the RAW to other color space (e.g. YUV) conversion. Since raw streams (e.g. RAW16) don't have the conversion stage, they are not croppable. The crop region will be ignored by raw streams.

For non-raw streams, any additional per-stream cropping will be done to maximize the final pixel area of the stream.

For example, if the crop region is set to a 4:3 aspect ratio, then 4:3 streams will use the exact crop region. 16:9 streams will further crop vertically (letterbox).

Conversely, if the crop region is set to a 16:9, then 4:3 outputs will crop horizontally (pillarbox), and 16:9 streams will match exactly. These additional crops will be centered within the crop region.

The width and height of the crop region cannot be set to be smaller than floor( activeArraySize.width / android.scaler.availableMaxDigitalZoom ) and floor( activeArraySize.height / android.scaler.availableMaxDigitalZoom ), respectively.

The camera device may adjust the crop region to account for rounding and other hardware requirements; the final crop region used will be included in the output capture result.

HAL Implementation Details

The output streams must maintain square pixels at all times, no matter what the relative aspect ratios of the crop region and the stream are. Negative values for corner are allowed for raw output if full pixel array is larger than active pixel array. Width and height may be rounded to nearest larger supportable width, especially for raw output, where only a few fixed scales may be possible.

For a set of output streams configured, if the sensor output is cropped to a smaller size than active array size, the HAL need follow below cropping rules:

  • The HAL need handle the cropRegion as if the sensor crop size is the effective active array size.More specifically, the HAL must transform the request cropRegion from android.sensor.info.activeArraySize to the sensor cropped pixel area size in this way:

    1. Translate the requested cropRegion w.r.t., the left top corner of the sensor cropped pixel area by (tx, ty), where tx = sensorCrop.top * (sensorCrop.height / activeArraySize.height) and tx = sensorCrop.left * (sensorCrop.width / activeArraySize.width). The (sensorCrop.top, sensorCrop.left) is the coordinate based off the android.sensor.info.activeArraySize.
    2. Scale the width and height of requested cropRegion with scaling factor of sensorCrop.width/activeArraySize.width and sensorCrop.height/activeArraySize.height respectively. Once this new cropRegion is calculated, the HAL must use this region to crop the image with regard to the sensor crop size (effective active array size). The HAL still need follow the general cropping rule for this new cropRegion and effective active array size.
  • The HAL must report the cropRegion with regard to android.sensor.info.activeArraySize. The HAL need convert the new cropRegion generated above w.r.t., full active array size. The reported cropRegion may be slightly different with the requested cropRegion since the HAL may adjust the crop region to account for rounding, conversion error, or other hardware limitations.

HAL2.x uses only (x, y, width)

sensor
controls
Property Name Type Description Units Range Tags
android.sensor.exposureTime int64 [public] [full]

Duration each pixel is exposed to light.

Nanoseconds

android.sensor.info.exposureTimeRange

Details

If the sensor can't expose this exact duration, it will shorten the duration exposed to the nearest possible value (rather than expose longer). The final exposure time used will be available in the output capture result.

This control is only effective if android.control.aeMode or android.control.mode is set to OFF; otherwise the auto-exposure algorithm will override this value.

android.sensor.frameDuration int64 [public] [full]

Duration from start of frame exposure to start of next frame exposure.

Nanoseconds

See android.sensor.info.maxFrameDuration, android.scaler.streamConfigurationMap. The duration is capped to max(duration, exposureTime + overhead).

Details

The maximum frame rate that can be supported by a camera subsystem is a function of many factors:

  • Requested resolutions of output image streams
  • Availability of binning / skipping modes on the imager
  • The bandwidth of the imager interface
  • The bandwidth of the various ISP processing blocks

Since these factors can vary greatly between different ISPs and sensors, the camera abstraction tries to represent the bandwidth restrictions with as simple a model as possible.

The model presented has the following characteristics:

  • The image sensor is always configured to output the smallest resolution possible given the application's requested output stream sizes. The smallest resolution is defined as being at least as large as the largest requested output stream size; the camera pipeline must never digitally upsample sensor data when the crop region covers the whole sensor. In general, this means that if only small output stream resolutions are configured, the sensor can provide a higher frame rate.
  • Since any request may use any or all the currently configured output streams, the sensor and ISP must be configured to support scaling a single capture to all the streams at the same time. This means the camera pipeline must be ready to produce the largest requested output size without any delay. Therefore, the overall frame rate of a given configured stream set is governed only by the largest requested stream resolution.
  • Using more than one output stream in a request does not affect the frame duration.
  • Certain format-streams may need to do additional background processing before data is consumed/produced by that stream. These processors can run concurrently to the rest of the camera pipeline, but cannot process more than 1 capture at a time.

The necessary information for the application, given the model above, is provided via the android.scaler.streamConfigurationMap field using StreamConfigurationMap#getOutputMinFrameDuration. These are used to determine the maximum frame rate / minimum frame duration that is possible for a given stream configuration.

Specifically, the application can use the following rules to determine the minimum frame duration it can request from the camera device:

  1. Let the set of currently configured input/output streams be called S.
  2. Find the minimum frame durations for each stream in S, by looking it up in android.scaler.streamConfigurationMap using StreamConfigurationMap#getOutputMinFrameDuration (with its respective size/format). Let this set of frame durations be called F.
  3. For any given request R, the minimum frame duration allowed for R is the maximum out of all values in F. Let the streams used in R be called S_r.

If none of the streams in S_r have a stall time (listed in StreamConfigurationMap#getOutputStallDuration using its respective size/format), then the frame duration in F determines the steady state frame rate that the application will get if it uses R as a repeating request. Let this special kind of request be called Rsimple.

A repeating request Rsimple can be occasionally interleaved by a single capture of a new request Rstall (which has at least one in-use stream with a non-0 stall time) and if Rstall has the same minimum frame duration this will not cause a frame rate loss if all buffers from the previous Rstall have already been delivered.

For more details about stalling, see StreamConfigurationMap#getOutputStallDuration.

This control is only effective if android.control.aeMode or android.control.mode is set to OFF; otherwise the auto-exposure algorithm will override this value.

HAL Implementation Details

For more details about stalling, see android.scaler.availableStallDurations.

android.sensor.sensitivity int32 [public] [full]

The amount of gain applied to sensor data before processing.

ISO arithmetic units

android.sensor.info.sensitivityRange

Details

The sensitivity is the standard ISO sensitivity value, as defined in ISO 12232:2006.

The sensitivity must be within android.sensor.info.sensitivityRange, and if if it less than android.sensor.maxAnalogSensitivity, the camera device is guaranteed to use only analog amplification for applying the gain.

If the camera device cannot apply the exact sensitivity requested, it will reduce the gain to the nearest supported value. The final sensitivity used will be available in the output capture result.

HAL Implementation Details

ISO 12232:2006 REI method is acceptable.

android.sensor.testPatternData int32 x 4 [public]

A pixel [R, G_even, G_odd, B] that supplies the test pattern when android.sensor.testPatternMode is SOLID_COLOR.

Details

Each color channel is treated as an unsigned 32-bit integer. The camera device then uses the most significant X bits that correspond to how many bits are in its Bayer raw sensor output.

For example, a sensor with RAW10 Bayer output would use the 10 most significant bits from each color channel.

HAL Implementation Details
android.sensor.testPatternMode int32 [public]
  • OFF

    No test pattern mode is used, and the camera device returns captures from the image sensor.

    This is the default if the key is not set.

  • SOLID_COLOR

    Each pixel in [R, G_even, G_odd, B] is replaced by its respective color channel provided in android.sensor.testPatternData.

    For example:

    android.testPatternData = [0, 0xFFFFFFFF, 0xFFFFFFFF, 0]
    

    All green pixels are 100% green. All red/blue pixels are black.

    android.testPatternData = [0xFFFFFFFF, 0, 0xFFFFFFFF, 0]
    

    All red pixels are 100% red. Only the odd green pixels are 100% green. All blue pixels are 100% black.

  • COLOR_BARS

    All pixel data is replaced with an 8-bar color pattern.

    The vertical bars (left-to-right) are as follows:

    • 100% white
    • yellow
    • cyan
    • green
    • magenta
    • red
    • blue
    • black

    In general the image would look like the following:

    W Y C G M R B K
    W Y C G M R B K
    W Y C G M R B K
    W Y C G M R B K
    W Y C G M R B K
    . . . . . . . .
    . . . . . . . .
    . . . . . . . .
    
    (B = Blue, K = Black)
    

    Each bar should take up 1/8 of the sensor pixel array width. When this is not possible, the bar size should be rounded down to the nearest integer and the pattern can repeat on the right side.

    Each bar's height must always take up the full sensor pixel array height.

    Each pixel in this test pattern must be set to either 0% intensity or 100% intensity.

  • COLOR_BARS_FADE_TO_GRAY

    The test pattern is similar to COLOR_BARS, except that each bar should start at its specified color at the top, and fade to gray at the bottom.

    Furthermore each bar is further subdivided into a left and right half. The left half should have a smooth gradient, and the right half should have a quantized gradient.

    In particular, the right half's should consist of blocks of the same color for 1/16th active sensor pixel array width.

    The least significant bits in the quantized gradient should be copied from the most significant bits of the smooth gradient.

    The height of each bar should always be a multiple of 128. When this is not the case, the pattern should repeat at the bottom of the image.

  • PN9

    All pixel data is replaced by a pseudo-random sequence generated from a PN9 512-bit sequence (typically implemented in hardware with a linear feedback shift register).

    The generator should be reset at the beginning of each frame, and thus each subsequent raw frame with this test pattern should be exactly the same as the last.

  • CUSTOM1 256

    The first custom test pattern. All custom patterns that are available only on this camera device are at least this numeric value.

    All of the custom test patterns will be static (that is the raw image must not vary from frame to frame).

When enabled, the sensor sends a test pattern instead of doing a real exposure from the camera.

android.sensor.availableTestPatternModes

Details

When a test pattern is enabled, all manual sensor controls specified by android.sensor.* will be ignored. All other controls should work as normal.

For example, if manual flash is enabled, flash firing should still occur (and that the test pattern remain unmodified, since the flash would not actually affect it).

Defaults to OFF.

HAL Implementation Details

All test patterns are specified in the Bayer domain.

The HAL may choose to substitute test patterns from the sensor with test patterns from on-device memory. In that case, it should be indistinguishable to the ISP whether the data came from the sensor interconnect bus (such as CSI2) or memory.

static
Property Name Type Description Units Range Tags
android.sensor.info.activeArraySize int32 x 4 [public as rectangle] [legacy]
Four ints defining the active pixel rectangle

The area of the image sensor which corresponds to active pixels after any geometric distortion correction has been applied.

Pixel coordinates on the image sensor
Details

This is the rectangle representing the size of the active region of the sensor (i.e. the region that actually receives light from the scene) after any geometric correction has been applied, and should be treated as the maximum size in pixels of any of the image output formats aside from the raw formats.

This rectangle is defined relative to the full pixel array; (0,0) is the top-left of the full pixel array, and the size of the full pixel array is given by android.sensor.info.pixelArraySize.

The coordinate system for most other keys that list pixel coordinates, including android.scaler.cropRegion, is defined relative to the active array rectangle given in this field, with (0, 0) being the top-left of this rectangle.

The active array may be smaller than the full pixel array, since the full array may include black calibration pixels or other inactive regions, and geometric correction resulting in scaling or cropping may have been applied.

HAL Implementation Details

This array contains (xmin, ymin, width, height). The (xmin, ymin) must be >= (0,0). The (width, height) must be <= android.sensor.info.pixelArraySize.

android.sensor.info.sensitivityRange int32 x 2 [public as rangeInt] [full]
Range of supported sensitivities

Range of sensitivities for android.sensor.sensitivity supported by this camera device.

Min <= 100, Max >= 800

Details

The values are the standard ISO sensitivity values, as defined in ISO 12232:2006.

android.sensor.info.colorFilterArrangement byte [public] [full]
  • RGGB
  • GRBG
  • GBRG
  • BGGR
  • RGB

    Sensor is not Bayer; output has 3 16-bit values for each pixel, instead of just 1 16-bit value per pixel.

The arrangement of color filters on sensor; represents the colors in the top-left 2x2 section of the sensor, in reading order.

android.sensor.info.exposureTimeRange int64 x 2 [public as rangeLong] [full]
nanoseconds

The range of image exposure times for android.sensor.exposureTime supported by this camera device.

Nanoseconds

The minimum exposure time will be less than 100 us. For FULL capability devices (android.info.supportedHardwareLevel == FULL), the maximum exposure time will be greater than 100ms.

HAL Implementation Details

For FULL capability devices (android.info.supportedHardwareLevel == FULL), The maximum of the range SHOULD be at least 1 second (1e9), MUST be at least 100ms.

android.sensor.info.maxFrameDuration int64 [public] [full]

The maximum possible frame duration (minimum frame rate) for android.sensor.frameDuration that is supported this camera device.

Nanoseconds

For FULL capability devices (android.info.supportedHardwareLevel == FULL), at least 100ms.

Details

Attempting to use frame durations beyond the maximum will result in the frame duration being clipped to the maximum. See that control for a full definition of frame durations.

Refer to StreamConfigurationMap#getOutputMinFrameDuration for the minimum frame duration values.

HAL Implementation Details

For FULL capability devices (android.info.supportedHardwareLevel == FULL), The maximum of the range SHOULD be at least 1 second (1e9), MUST be at least 100ms (100e6).

android.sensor.info.maxFrameDuration must be greater or equal to the android.sensor.info.exposureTimeRange max value (since exposure time overrides frame duration).

Available minimum frame durations for JPEG must be no greater than that of the YUV_420_888/IMPLEMENTATION_DEFINED minimum frame durations (for that respective size).

Since JPEG processing is considered offline and can take longer than a single uncompressed capture, refer to android.scaler.availableStallDurations for details about encoding this scenario.

android.sensor.info.physicalSize float x 2 [public as sizeF] [legacy]
width x height

The physical dimensions of the full pixel array.

Millimeters
Details

This is the physical size of the sensor pixel array defined by android.sensor.info.pixelArraySize.

HAL Implementation Details

Needed for FOV calculation for old API

android.sensor.info.pixelArraySize int32 x 2 [public as size] [legacy]

Dimensions of the full pixel array, possibly including black calibration pixels.

Pixels
Details

The pixel count of the full pixel array of the image sensor, which covers android.sensor.info.physicalSize area. This represents the full pixel dimensions of the raw buffers produced by this sensor.

If a camera device supports raw sensor formats, either this or android.sensor.info.preCorrectionActiveArraySize is the maximum dimensions for the raw output formats listed in android.scaler.streamConfigurationMap (this depends on whether or not the image sensor returns buffers containing pixels that are not part of the active array region for blacklevel calibration or other purposes).

Some parts of the full pixel array may not receive light from the scene, or be otherwise inactive. The android.sensor.info.preCorrectionActiveArraySize key defines the rectangle of active pixels that will be included in processed image formats.

android.sensor.info.whiteLevel int32 [public]

Maximum raw value output by sensor.

> 255 (8-bit output)

Details

This specifies the fully-saturated encoding level for the raw sample values from the sensor. This is typically caused by the sensor becoming highly non-linear or clipping. The minimum for each channel is specified by the offset in the android.sensor.blackLevelPattern key.

The white level is typically determined either by sensor bit depth (8-14 bits is expected), or by the point where the sensor response becomes too non-linear to be useful. The default value for this is maximum representable value for a 16-bit raw sample (2^16 - 1).

HAL Implementation Details

The full bit depth of the sensor must be available in the raw data, so the value for linear sensors should not be significantly lower than maximum raw value supported, i.e. 2^(sensor bits per pixel).

android.sensor.info.timestampSource byte [public] [legacy]
  • UNKNOWN

    Timestamps from android.sensor.timestamp are in nanoseconds and monotonic, but can not be compared to timestamps from other subsystems (e.g. accelerometer, gyro etc.), or other instances of the same or different camera devices in the same system. Timestamps between streams and results for a single camera instance are comparable, and the timestamps for all buffers and the result metadata generated by a single capture are identical.

  • REALTIME

    Timestamps from android.sensor.timestamp are in the same timebase as SystemClock#elapsedRealtimeNanos, and they can be compared to other timestamps using that base.

The time base source for sensor capture start timestamps.

Details

The timestamps provided for captures are always in nanoseconds and monotonic, but may not based on a time source that can be compared to other system time sources.

This characteristic defines the source for the timestamps, and therefore whether they can be compared against other system time sources/timestamps.

android.sensor.info.lensShadingApplied byte [public as boolean]
  • FALSE
  • TRUE

Whether the RAW images output from this camera device are subject to lens shading correction.

Details

If TRUE, all images produced by the camera device in the RAW image formats will have lens shading correction already applied to it. If FALSE, the images will not be adjusted for lens shading correction. See android.request.maxNumOutputRaw for a list of RAW image formats.

This key will be null for all devices do not report this information. Devices with RAW capability will always report this information in this key.

android.sensor.info.preCorrectionActiveArraySize int32 x 4 [public as rectangle] [legacy]
Four ints defining the active pixel rectangle

The area of the image sensor which corresponds to active pixels prior to the application of any geometric distortion correction.

Pixel coordinates on the image sensor
Details

This is the rectangle representing the size of the active region of the sensor (i.e. the region that actually receives light from the scene) before any geometric correction has been applied, and should be treated as the active region rectangle for any of the raw formats. All metadata associated with raw processing (e.g. the lens shading correction map, and radial distortion fields) treats the top, left of this rectangle as the origin, (0,0).

The size of this region determines the maximum field of view and the maximum number of pixels that an image from this sensor can contain, prior to the application of geometric distortion correction. The effective maximum pixel dimensions of a post-distortion-corrected image is given by the android.sensor.info.activeArraySize field, and the effective maximum field of view for a post-distortion-corrected image can be calculated by applying the geometric distortion correction fields to this rectangle, and cropping to the rectangle given in android.sensor.info.activeArraySize.

E.g. to calculate position of a pixel, (x,y), in a processed YUV output image with the dimensions in android.sensor.info.activeArraySize given the position of a pixel, (x', y'), in the raw pixel array with dimensions give in android.sensor.info.pixelArraySize:

  1. Choose a pixel (x', y') within the active array region of the raw buffer given in android.sensor.info.preCorrectionActiveArraySize, otherwise this pixel is considered to be outside of the FOV, and will not be shown in the processed output image.
  2. Apply geometric distortion correction to get the post-distortion pixel coordinate, (x_i, y_i). When applying geometric correction metadata, note that metadata for raw buffers is defined relative to the top, left of the android.sensor.info.preCorrectionActiveArraySize rectangle.
  3. If the resulting corrected pixel coordinate is within the region given in android.sensor.info.activeArraySize, then the position of this pixel in the processed output image buffer is (x_i - activeArray.left, y_i - activeArray.top), when the top, left coordinate of that buffer is treated as (0, 0).

Thus, for pixel x',y' = (25, 25) on a sensor where android.sensor.info.pixelArraySize is (100,100), android.sensor.info.preCorrectionActiveArraySize is (10, 10, 100, 100), android.sensor.info.activeArraySize is (20, 20, 80, 80), and the geometric distortion correction doesn't change the pixel coordinate, the resulting pixel selected in pixel coordinates would be x,y = (25, 25) relative to the top,left of the raw buffer with dimensions given in android.sensor.info.pixelArraySize, and would be (5, 5) relative to the top,left of post-processed YUV output buffer with dimensions given in android.sensor.info.activeArraySize.

The currently supported fields that correct for geometric distortion are:

  1. android.lens.radialDistortion.

If all of the geometric distortion fields are no-ops, this rectangle will be the same as the post-distortion-corrected rectangle given in android.sensor.info.activeArraySize.

This rectangle is defined relative to the full pixel array; (0,0) is the top-left of the full pixel array, and the size of the full pixel array is given by android.sensor.info.pixelArraySize.

The pre-correction active array may be smaller than the full pixel array, since the full array may include black calibration pixels or other inactive regions.

HAL Implementation Details

This array contains (xmin, ymin, width, height). The (xmin, ymin) must be >= (0,0). The (width, height) must be <= android.sensor.info.pixelArraySize.

If omitted by the HAL implementation, the camera framework will assume that this is the same as the post-correction active array region given in android.sensor.info.activeArraySize.

android.sensor.referenceIlluminant1 byte [public]
  • DAYLIGHT 1
  • FLUORESCENT 2
  • TUNGSTEN 3

    Incandescent light

  • FLASH 4
  • FINE_WEATHER 9
  • CLOUDY_WEATHER 10
  • SHADE 11
  • DAYLIGHT_FLUORESCENT 12

    D 5700 - 7100K

  • DAY_WHITE_FLUORESCENT 13

    N 4600 - 5400K

  • COOL_WHITE_FLUORESCENT 14

    W 3900 - 4500K

  • WHITE_FLUORESCENT 15

    WW 3200 - 3700K

  • STANDARD_A 17
  • STANDARD_B 18
  • STANDARD_C 19
  • D55 20
  • D65 21
  • D75 22
  • D50 23
  • ISO_STUDIO_TUNGSTEN 24

The standard reference illuminant used as the scene light source when calculating the android.sensor.colorTransform1, android.sensor.calibrationTransform1, and android.sensor.forwardMatrix1 matrices.

Details

The values in this key correspond to the values defined for the EXIF LightSource tag. These illuminants are standard light sources that are often used calibrating camera devices.

If this key is present, then android.sensor.colorTransform1, android.sensor.calibrationTransform1, and android.sensor.forwardMatrix1 will also be present.

Some devices may choose to provide a second set of calibration information for improved quality, including android.sensor.referenceIlluminant2 and its corresponding matrices.

HAL Implementation Details

The first reference illuminant (android.sensor.referenceIlluminant1) and corresponding matrices must be present to support the RAW capability and DNG output.

When producing raw images with a color profile that has only been calibrated against a single light source, it is valid to omit android.sensor.referenceIlluminant2 along with the android.sensor.colorTransform2, android.sensor.calibrationTransform2, and android.sensor.forwardMatrix2 matrices.

If only android.sensor.referenceIlluminant1 is included, it should be chosen so that it is representative of typical scene lighting. In general, D50 or DAYLIGHT will be chosen for this case.

If both android.sensor.referenceIlluminant1 and android.sensor.referenceIlluminant2 are included, they should be chosen to represent the typical range of scene lighting conditions. In general, low color temperature illuminant such as Standard-A will be chosen for the first reference illuminant and a higher color temperature illuminant such as D65 will be chosen for the second reference illuminant.

android.sensor.referenceIlluminant2 byte [public]

The standard reference illuminant used as the scene light source when calculating the android.sensor.colorTransform2, android.sensor.calibrationTransform2, and android.sensor.forwardMatrix2 matrices.

Any value listed in android.sensor.referenceIlluminant1

Details

See android.sensor.referenceIlluminant1 for more details.

If this key is present, then android.sensor.colorTransform2, android.sensor.calibrationTransform2, and android.sensor.forwardMatrix2 will also be present.

android.sensor.calibrationTransform1 rational x 3 x 3 [public as colorSpaceTransform]
3x3 matrix in row-major-order

A per-device calibration transform matrix that maps from the reference sensor colorspace to the actual device sensor colorspace.

Details

This matrix is used to correct for per-device variations in the sensor colorspace, and is used for processing raw buffer data.

The matrix is expressed as a 3x3 matrix in row-major-order, and contains a per-device calibration transform that maps colors from reference sensor color space (i.e. the "golden module" colorspace) into this camera device's native sensor color space under the first reference illuminant (android.sensor.referenceIlluminant1).

android.sensor.calibrationTransform2 rational x 3 x 3 [public as colorSpaceTransform]
3x3 matrix in row-major-order

A per-device calibration transform matrix that maps from the reference sensor colorspace to the actual device sensor colorspace (this is the colorspace of the raw buffer data).

Details

This matrix is used to correct for per-device variations in the sensor colorspace, and is used for processing raw buffer data.

The matrix is expressed as a 3x3 matrix in row-major-order, and contains a per-device calibration transform that maps colors from reference sensor color space (i.e. the "golden module" colorspace) into this camera device's native sensor color space under the second reference illuminant (android.sensor.referenceIlluminant2).

This matrix will only be present if the second reference illuminant is present.

android.sensor.colorTransform1 rational x 3 x 3 [public as colorSpaceTransform]
3x3 matrix in row-major-order

A matrix that transforms color values from CIE XYZ color space to reference sensor color space.

Details

This matrix is used to convert from the standard CIE XYZ color space to the reference sensor colorspace, and is used when processing raw buffer data.

The matrix is expressed as a 3x3 matrix in row-major-order, and contains a color transform matrix that maps colors from the CIE XYZ color space to the reference sensor color space (i.e. the "golden module" colorspace) under the first reference illuminant (android.sensor.referenceIlluminant1).

The white points chosen in both the reference sensor color space and the CIE XYZ colorspace when calculating this transform will match the standard white point for the first reference illuminant (i.e. no chromatic adaptation will be applied by this transform).

android.sensor.colorTransform2 rational x 3 x 3 [public as colorSpaceTransform]
3x3 matrix in row-major-order

A matrix that transforms color values from CIE XYZ color space to reference sensor color space.

Details

This matrix is used to convert from the standard CIE XYZ color space to the reference sensor colorspace, and is used when processing raw buffer data.

The matrix is expressed as a 3x3 matrix in row-major-order, and contains a color transform matrix that maps colors from the CIE XYZ color space to the reference sensor color space (i.e. the "golden module" colorspace) under the second reference illuminant (android.sensor.referenceIlluminant2).

The white points chosen in both the reference sensor color space and the CIE XYZ colorspace when calculating this transform will match the standard white point for the second reference illuminant (i.e. no chromatic adaptation will be applied by this transform).

This matrix will only be present if the second reference illuminant is present.

android.sensor.forwardMatrix1 rational x 3 x 3 [public as colorSpaceTransform]
3x3 matrix in row-major-order

A matrix that transforms white balanced camera colors from the reference sensor colorspace to the CIE XYZ colorspace with a D50 whitepoint.

Details

This matrix is used to convert to the standard CIE XYZ colorspace, and is used when processing raw buffer data.

This matrix is expressed as a 3x3 matrix in row-major-order, and contains a color transform matrix that maps white balanced colors from the reference sensor color space to the CIE XYZ color space with a D50 white point.

Under the first reference illuminant (android.sensor.referenceIlluminant1) this matrix is chosen so that the standard white point for this reference illuminant in the reference sensor colorspace is mapped to D50 in the CIE XYZ colorspace.

android.sensor.forwardMatrix2 rational x 3 x 3 [public as colorSpaceTransform]
3x3 matrix in row-major-order

A matrix that transforms white balanced camera colors from the reference sensor colorspace to the CIE XYZ colorspace with a D50 whitepoint.

Details

This matrix is used to convert to the standard CIE XYZ colorspace, and is used when processing raw buffer data.

This matrix is expressed as a 3x3 matrix in row-major-order, and contains a color transform matrix that maps white balanced colors from the reference sensor color space to the CIE XYZ color space with a D50 white point.

Under the second reference illuminant (android.sensor.referenceIlluminant2) this matrix is chosen so that the standard white point for this reference illuminant in the reference sensor colorspace is mapped to D50 in the CIE XYZ colorspace.

This matrix will only be present if the second reference illuminant is present.

android.sensor.baseGainFactor rational [system]

Gain factor from electrons to raw units when ISO=100

android.sensor.blackLevelPattern int32 x 4 [public as blackLevelPattern]
2x2 raw count block

A fixed black level offset for each of the color filter arrangement (CFA) mosaic channels.

>= 0 for each.

Details

This key specifies the zero light value for each of the CFA mosaic channels in the camera sensor. The maximal value output by the sensor is represented by the value in android.sensor.info.whiteLevel.

The values are given in the same order as channels listed for the CFA layout key (see android.sensor.info.colorFilterArrangement), i.e. the nth value given corresponds to the black level offset for the nth color channel listed in the CFA.

HAL Implementation Details

The values are given in row-column scan order, with the first value corresponding to the element of the CFA in row=0, column=0.

android.sensor.maxAnalogSensitivity int32 [public] [full]

Maximum sensitivity that is implemented purely through analog gain.

Details

For android.sensor.sensitivity values less than or equal to this, all applied gain must be analog. For values above this, the gain applied can be a mix of analog and digital.

android.sensor.orientation int32 [public] [legacy]

Clockwise angle through which the output image needs to be rotated to be upright on the device screen in its native orientation.

Degrees of clockwise rotation; always a multiple of 90

0, 90, 180, 270

Details

Also defines the direction of rolling shutter readout, which is from top to bottom in the sensor's coordinate system.

android.sensor.profileHueSatMapDimensions int32 x 3 [system]
Number of samples for hue, saturation, and value

The number of input samples for each dimension of android.sensor.profileHueSatMap.

Hue >= 1, Saturation >= 2, Value >= 1

Details

The number of input samples for the hue, saturation, and value dimension of android.sensor.profileHueSatMap. The order of the dimensions given is hue, saturation, value; where hue is the 0th element.

android.sensor.availableTestPatternModes int32 x n [public]
list of enums

List of sensor test pattern modes for android.sensor.testPatternMode supported by this camera device.

Any value listed in android.sensor.testPatternMode

Details

Defaults to OFF, and always includes OFF if defined.

HAL Implementation Details

All custom modes must be >= CUSTOM1.

dynamic
Property Name Type Description Units Range Tags
android.sensor.exposureTime int64 [public] [full]

Duration each pixel is exposed to light.

Nanoseconds

android.sensor.info.exposureTimeRange

Details

If the sensor can't expose this exact duration, it will shorten the duration exposed to the nearest possible value (rather than expose longer). The final exposure time used will be available in the output capture result.

This control is only effective if android.control.aeMode or android.control.mode is set to OFF; otherwise the auto-exposure algorithm will override this value.

android.sensor.frameDuration int64 [public] [full]

Duration from start of frame exposure to start of next frame exposure.

Nanoseconds

See android.sensor.info.maxFrameDuration, android.scaler.streamConfigurationMap. The duration is capped to max(duration, exposureTime + overhead).

Details

The maximum frame rate that can be supported by a camera subsystem is a function of many factors:

  • Requested resolutions of output image streams
  • Availability of binning / skipping modes on the imager
  • The bandwidth of the imager interface
  • The bandwidth of the various ISP processing blocks

Since these factors can vary greatly between different ISPs and sensors, the camera abstraction tries to represent the bandwidth restrictions with as simple a model as possible.

The model presented has the following characteristics:

  • The image sensor is always configured to output the smallest resolution possible given the application's requested output stream sizes. The smallest resolution is defined as being at least as large as the largest requested output stream size; the camera pipeline must never digitally upsample sensor data when the crop region covers the whole sensor. In general, this means that if only small output stream resolutions are configured, the sensor can provide a higher frame rate.
  • Since any request may use any or all the currently configured output streams, the sensor and ISP must be configured to support scaling a single capture to all the streams at the same time. This means the camera pipeline must be ready to produce the largest requested output size without any delay. Therefore, the overall frame rate of a given configured stream set is governed only by the largest requested stream resolution.
  • Using more than one output stream in a request does not affect the frame duration.
  • Certain format-streams may need to do additional background processing before data is consumed/produced by that stream. These processors can run concurrently to the rest of the camera pipeline, but cannot process more than 1 capture at a time.

The necessary information for the application, given the model above, is provided via the android.scaler.streamConfigurationMap field using StreamConfigurationMap#getOutputMinFrameDuration. These are used to determine the maximum frame rate / minimum frame duration that is possible for a given stream configuration.

Specifically, the application can use the following rules to determine the minimum frame duration it can request from the camera device:

  1. Let the set of currently configured input/output streams be called S.
  2. Find the minimum frame durations for each stream in S, by looking it up in android.scaler.streamConfigurationMap using StreamConfigurationMap#getOutputMinFrameDuration (with its respective size/format). Let this set of frame durations be called F.
  3. For any given request R, the minimum frame duration allowed for R is the maximum out of all values in F. Let the streams used in R be called S_r.

If none of the streams in S_r have a stall time (listed in StreamConfigurationMap#getOutputStallDuration using its respective size/format), then the frame duration in F determines the steady state frame rate that the application will get if it uses R as a repeating request. Let this special kind of request be called Rsimple.

A repeating request Rsimple can be occasionally interleaved by a single capture of a new request Rstall (which has at least one in-use stream with a non-0 stall time) and if Rstall has the same minimum frame duration this will not cause a frame rate loss if all buffers from the previous Rstall have already been delivered.

For more details about stalling, see StreamConfigurationMap#getOutputStallDuration.

This control is only effective if android.control.aeMode or android.control.mode is set to OFF; otherwise the auto-exposure algorithm will override this value.

HAL Implementation Details

For more details about stalling, see android.scaler.availableStallDurations.

android.sensor.sensitivity int32 [public] [full]

The amount of gain applied to sensor data before processing.

ISO arithmetic units

android.sensor.info.sensitivityRange

Details

The sensitivity is the standard ISO sensitivity value, as defined in ISO 12232:2006.

The sensitivity must be within android.sensor.info.sensitivityRange, and if if it less than android.sensor.maxAnalogSensitivity, the camera device is guaranteed to use only analog amplification for applying the gain.

If the camera device cannot apply the exact sensitivity requested, it will reduce the gain to the nearest supported value. The final sensitivity used will be available in the output capture result.

HAL Implementation Details

ISO 12232:2006 REI method is acceptable.

android.sensor.timestamp int64 [public] [legacy]

Time at start of exposure of first row of the image sensor active array, in nanoseconds.

Nanoseconds

> 0

Details

The timestamps are also included in all image buffers produced for the same capture, and will be identical on all the outputs.

When android.sensor.info.timestampSource == UNKNOWN, the timestamps measure time since an unspecified starting point, and are monotonically increasing. They can be compared with the timestamps for other captures from the same camera device, but are not guaranteed to be comparable to any other time source.

When android.sensor.info.timestampSource == REALTIME, the timestamps measure time in the same timebase as SystemClock#elapsedRealtimeNanos, and they can be compared to other timestamps from other subsystems that are using that base.

For reprocessing, the timestamp will match the start of exposure of the input image, i.e. the timestamp in the TotalCaptureResult that was used to create the reprocess capture request.

HAL Implementation Details

All timestamps must be in reference to the kernel's CLOCK_BOOTTIME monotonic clock, which properly accounts for time spent asleep. This allows for synchronization with sensors that continue to operate while the system is otherwise asleep.

If android.sensor.info.timestampSource == REALTIME, The timestamp must be synchronized with the timestamps from other sensor subsystems that are using the same timebase.

For reprocessing, the input image's start of exposure can be looked up with android.sensor.timestamp from the metadata included in the capture request.

android.sensor.temperature float [system]

The temperature of the sensor, sampled at the time exposure began for this frame.

The thermal diode being queried should be inside the sensor PCB, or somewhere close to it.

Celsius

Optional. This value is missing if no temperature is available.

android.sensor.neutralColorPoint rational x 3 [public]

The estimated camera neutral color in the native sensor colorspace at the time of capture.

Details

This value gives the neutral color point encoded as an RGB value in the native sensor color space. The neutral color point indicates the currently estimated white point of the scene illumination. It can be used to interpolate between the provided color transforms when processing raw sensor data.

The order of the values is R, G, B; where R is in the lowest index.

android.sensor.noiseProfile double x 2 x CFA Channels [public as pairDoubleDouble]
Pairs of noise model coefficients

Noise model coefficients for each CFA mosaic channel.

Details

This key contains two noise model coefficients for each CFA channel corresponding to the sensor amplification (S) and sensor readout noise (O). These are given as pairs of coefficients for each channel in the same order as channels listed for the CFA layout key (see android.sensor.info.colorFilterArrangement). This is represented as an array of Pair<Double, Double>, where the first member of the Pair at index n is the S coefficient and the second member is the O coefficient for the nth color channel in the CFA.

These coefficients are used in a two parameter noise model to describe the amount of noise present in the image for each CFA channel. The noise model used here is:

N(x) = sqrt(Sx + O)

Where x represents the recorded signal of a CFA channel normalized to the range [0, 1], and S and O are the noise model coeffiecients for that channel.

A more detailed description of the noise model can be found in the Adobe DNG specification for the NoiseProfile tag.

HAL Implementation Details

For a CFA layout of RGGB, the list of coefficients would be given as an array of doubles S0,O0,S1,O1,..., where S0 and O0 are the coefficients for the red channel, S1 and O1 are the coefficients for the first green channel, etc.

android.sensor.profileHueSatMap float x hue_samples x saturation_samples x value_samples x 3 [system]
Mapping for hue, saturation, and value

A mapping containing a hue shift, saturation scale, and value scale for each pixel.

The hue shift is given in degrees; saturation and value scale factors are unitless and are between 0 and 1 inclusive
Details

hue_samples, saturation_samples, and value_samples are given in android.sensor.profileHueSatMapDimensions.

Each entry of this map contains three floats corresponding to the hue shift, saturation scale, and value scale, respectively; where the hue shift has the lowest index. The map entries are stored in the key in nested loop order, with the value divisions in the outer loop, the hue divisions in the middle loop, and the saturation divisions in the inner loop. All zero input saturation entries are required to have a value scale factor of 1.0.

android.sensor.profileToneCurve float x samples x 2 [system]
Samples defining a spline for a tone-mapping curve

A list of x,y samples defining a tone-mapping curve for gamma adjustment.

Each sample has an input range of [0, 1] and an output range of [0, 1]. The first sample is required to be (0, 0), and the last sample is required to be (1, 1).

Details

This key contains a default tone curve that can be applied while processing the image as a starting point for user adjustments. The curve is specified as a list of value pairs in linear gamma. The curve is interpolated using a cubic spline.

android.sensor.greenSplit float [public]

The worst-case divergence between Bayer green channels.

>= 0

Details

This value is an estimate of the worst case split between the Bayer green channels in the red and blue rows in the sensor color filter array.

The green split is calculated as follows:

  1. A 5x5 pixel (or larger) window W within the active sensor array is chosen. The term 'pixel' here is taken to mean a group of 4 Bayer mosaic channels (R, Gr, Gb, B). The location and size of the window chosen is implementation defined, and should be chosen to provide a green split estimate that is both representative of the entire image for this camera sensor, and can be calculated quickly.
  2. The arithmetic mean of the green channels from the red rows (mean_Gr) within W is computed.
  3. The arithmetic mean of the green channels from the blue rows (mean_Gb) within W is computed.
  4. The maximum ratio R of the two means is computed as follows: R = max((mean_Gr + 1)/(mean_Gb + 1), (mean_Gb + 1)/(mean_Gr + 1))

The ratio R is the green split divergence reported for this property, which represents how much the green channels differ in the mosaic pattern. This value is typically used to determine the treatment of the green mosaic channels when demosaicing.

The green split value can be roughly interpreted as follows:

  • R < 1.03 is a negligible split (<3% divergence).
  • 1.20 <= R >= 1.03 will require some software correction to avoid demosaic errors (3-20% divergence).
  • R > 1.20 will require strong software correction to produce a usuable image (>20% divergence).
HAL Implementation Details

The green split given may be a static value based on prior characterization of the camera sensor using the green split calculation method given here over a large, representative, sample set of images. Other methods of calculation that produce equivalent results, and can be interpreted in the same manner, may be used.

android.sensor.testPatternData int32 x 4 [public]

A pixel [R, G_even, G_odd, B] that supplies the test pattern when android.sensor.testPatternMode is SOLID_COLOR.

Details

Each color channel is treated as an unsigned 32-bit integer. The camera device then uses the most significant X bits that correspond to how many bits are in its Bayer raw sensor output.

For example, a sensor with RAW10 Bayer output would use the 10 most significant bits from each color channel.

HAL Implementation Details
android.sensor.testPatternMode int32 [public]
  • OFF

    No test pattern mode is used, and the camera device returns captures from the image sensor.

    This is the default if the key is not set.

  • SOLID_COLOR

    Each pixel in [R, G_even, G_odd, B] is replaced by its respective color channel provided in android.sensor.testPatternData.

    For example:

    android.testPatternData = [0, 0xFFFFFFFF, 0xFFFFFFFF, 0]
    

    All green pixels are 100% green. All red/blue pixels are black.

    android.testPatternData = [0xFFFFFFFF, 0, 0xFFFFFFFF, 0]
    

    All red pixels are 100% red. Only the odd green pixels are 100% green. All blue pixels are 100% black.

  • COLOR_BARS

    All pixel data is replaced with an 8-bar color pattern.

    The vertical bars (left-to-right) are as follows:

    • 100% white
    • yellow
    • cyan
    • green
    • magenta
    • red
    • blue
    • black

    In general the image would look like the following:

    W Y C G M R B K
    W Y C G M R B K
    W Y C G M R B K
    W Y C G M R B K
    W Y C G M R B K
    . . . . . . . .
    . . . . . . . .
    . . . . . . . .
    
    (B = Blue, K = Black)
    

    Each bar should take up 1/8 of the sensor pixel array width. When this is not possible, the bar size should be rounded down to the nearest integer and the pattern can repeat on the right side.

    Each bar's height must always take up the full sensor pixel array height.

    Each pixel in this test pattern must be set to either 0% intensity or 100% intensity.

  • COLOR_BARS_FADE_TO_GRAY

    The test pattern is similar to COLOR_BARS, except that each bar should start at its specified color at the top, and fade to gray at the bottom.

    Furthermore each bar is further subdivided into a left and right half. The left half should have a smooth gradient, and the right half should have a quantized gradient.

    In particular, the right half's should consist of blocks of the same color for 1/16th active sensor pixel array width.

    The least significant bits in the quantized gradient should be copied from the most significant bits of the smooth gradient.

    The height of each bar should always be a multiple of 128. When this is not the case, the pattern should repeat at the bottom of the image.

  • PN9

    All pixel data is replaced by a pseudo-random sequence generated from a PN9 512-bit sequence (typically implemented in hardware with a linear feedback shift register).

    The generator should be reset at the beginning of each frame, and thus each subsequent raw frame with this test pattern should be exactly the same as the last.

  • CUSTOM1 256

    The first custom test pattern. All custom patterns that are available only on this camera device are at least this numeric value.

    All of the custom test patterns will be static (that is the raw image must not vary from frame to frame).

When enabled, the sensor sends a test pattern instead of doing a real exposure from the camera.

android.sensor.availableTestPatternModes

Details

When a test pattern is enabled, all manual sensor controls specified by android.sensor.* will be ignored. All other controls should work as normal.

For example, if manual flash is enabled, flash firing should still occur (and that the test pattern remain unmodified, since the flash would not actually affect it).

Defaults to OFF.

HAL Implementation Details

All test patterns are specified in the Bayer domain.

The HAL may choose to substitute test patterns from the sensor with test patterns from on-device memory. In that case, it should be indistinguishable to the ISP whether the data came from the sensor interconnect bus (such as CSI2) or memory.

android.sensor.rollingShutterSkew int64 [public] [limited]

Duration between the start of first row exposure and the start of last row exposure.

Nanoseconds

>= 0 and < StreamConfigurationMap#getOutputMinFrameDuration.

Details

This is the exposure time skew between the first and last row exposure start times. The first row and the last row are the first and last rows inside of the android.sensor.info.activeArraySize.

For typical camera sensors that use rolling shutters, this is also equivalent to the frame readout time.

HAL Implementation Details

The HAL must report 0 if the sensor is using global shutter, where all pixels begin exposure at the same time.

shading
controls
Property Name Type Description Units Range Tags
android.shading.mode byte [public] [full]
  • OFF

    No lens shading correction is applied.

  • FAST

    Apply lens shading corrections, without slowing frame rate relative to sensor raw output

  • HIGH_QUALITY

    Apply high-quality lens shading correction, at the cost of possibly reduced frame rate.

Quality of lens shading correction applied to the image data.

android.shading.availableModes

Details

When set to OFF mode, no lens shading correction will be applied by the camera device, and an identity lens shading map data will be provided if android.statistics.lensShadingMapMode == ON. For example, for lens shading map with size of [ 4, 3 ], the output android.statistics.lensShadingCorrectionMap for this case will be an identity map shown below:

[ 1.0, 1.0, 1.0, 1.0,  1.0, 1.0, 1.0, 1.0,
 1.0, 1.0, 1.0, 1.0,  1.0, 1.0, 1.0, 1.0,
 1.0, 1.0, 1.0, 1.0,  1.0, 1.0, 1.0, 1.0,
 1.0, 1.0, 1.0, 1.0,  1.0, 1.0, 1.0, 1.0,
 1.0, 1.0, 1.0, 1.0,  1.0, 1.0, 1.0, 1.0,
 1.0, 1.0, 1.0, 1.0,  1.0, 1.0, 1.0, 1.0 ]

When set to other modes, lens shading correction will be applied by the camera device. Applications can request lens shading map data by setting android.statistics.lensShadingMapMode to ON, and then the camera device will provide lens shading map data in android.statistics.lensShadingCorrectionMap; the returned shading map data will be the one applied by the camera device for this capture request.

The shading map data may depend on the auto-exposure (AE) and AWB statistics, therefore the reliability of the map data may be affected by the AE and AWB algorithms. When AE and AWB are in AUTO modes(android.control.aeMode != OFF and android.control.awbMode != OFF), to get best results, it is recommended that the applications wait for the AE and AWB to be converged before using the returned shading map data.

android.shading.strength byte [system]

Control the amount of shading correction applied to the images

unitless: 1-10; 10 is full shading compensation
dynamic
Property Name Type Description Units Range Tags
android.shading.mode byte [public] [full]
  • OFF

    No lens shading correction is applied.

  • FAST

    Apply lens shading corrections, without slowing frame rate relative to sensor raw output

  • HIGH_QUALITY

    Apply high-quality lens shading correction, at the cost of possibly reduced frame rate.

Quality of lens shading correction applied to the image data.

android.shading.availableModes

Details

When set to OFF mode, no lens shading correction will be applied by the camera device, and an identity lens shading map data will be provided if android.statistics.lensShadingMapMode == ON. For example, for lens shading map with size of [ 4, 3 ], the output android.statistics.lensShadingCorrectionMap for this case will be an identity map shown below:

[ 1.0, 1.0, 1.0, 1.0,  1.0, 1.0, 1.0, 1.0,
 1.0, 1.0, 1.0, 1.0,  1.0, 1.0, 1.0, 1.0,
 1.0, 1.0, 1.0, 1.0,  1.0, 1.0, 1.0, 1.0,
 1.0, 1.0, 1.0, 1.0,  1.0, 1.0, 1.0, 1.0,
 1.0, 1.0, 1.0, 1.0,  1.0, 1.0, 1.0, 1.0,
 1.0, 1.0, 1.0, 1.0,  1.0, 1.0, 1.0, 1.0 ]

When set to other modes, lens shading correction will be applied by the camera device. Applications can request lens shading map data by setting android.statistics.lensShadingMapMode to ON, and then the camera device will provide lens shading map data in android.statistics.lensShadingCorrectionMap; the returned shading map data will be the one applied by the camera device for this capture request.

The shading map data may depend on the auto-exposure (AE) and AWB statistics, therefore the reliability of the map data may be affected by the AE and AWB algorithms. When AE and AWB are in AUTO modes(android.control.aeMode != OFF and android.control.awbMode != OFF), to get best results, it is recommended that the applications wait for the AE and AWB to be converged before using the returned shading map data.

static
Property Name Type Description Units Range Tags
android.shading.availableModes byte x n [public as enumList] [legacy]
List of enums (android.shading.mode).

List of lens shading modes for android.shading.mode that are supported by this camera device.

Any value listed in android.shading.mode

Details

This list contains lens shading modes that can be set for the camera device. Camera devices that support the MANUAL_POST_PROCESSING capability will always list OFF and FAST mode. This includes all FULL level devices. LEGACY devices will always only support FAST mode.

HAL Implementation Details

HAL must support both FAST and HIGH_QUALITY if lens shading correction control is available on the camera device, but the underlying implementation can be the same for both modes. That is, if the highest quality implementation on the camera device does not slow down capture rate, then FAST and HIGH_QUALITY will generate the same output.

statistics
controls
Property Name Type Description Units Range Tags
android.statistics.faceDetectMode byte [public] [legacy]
  • OFF

    Do not include face detection statistics in capture results.

  • SIMPLE [optional]

    Return face rectangle and confidence values only.

  • FULL [optional]

    Return all face metadata.

    In this mode, face rectangles, scores, landmarks, and face IDs are all valid.

Operating mode for the face detector unit.

android.statistics.info.availableFaceDetectModes

Details

Whether face detection is enabled, and whether it should output just the basic fields or the full set of fields.

HAL Implementation Details

SIMPLE mode must fill in android.statistics.faceRectangles and android.statistics.faceScores. FULL mode must also fill in android.statistics.faceIds, and android.statistics.faceLandmarks.

android.statistics.histogramMode byte [system as boolean]
  • OFF
  • ON

Operating mode for histogram generation

android.statistics.sharpnessMapMode byte [system as boolean]
  • OFF
  • ON

Operating mode for sharpness map generation

android.statistics.hotPixelMapMode byte [public as boolean]
  • OFF

    Hot pixel map production is disabled.

  • ON

    Hot pixel map production is enabled.

Operating mode for hot pixel map generation.

android.statistics.info.availableHotPixelMapModes

Details

If set to true, a hot pixel map is returned in android.statistics.hotPixelMap. If set to false, no hot pixel map will be returned.

android.statistics.lensShadingMapMode byte [public] [full]
  • OFF

    Do not include a lens shading map in the capture result.

  • ON

    Include a lens shading map in the capture result.

Whether the camera device will output the lens shading map in output result metadata.

android.statistics.info.availableLensShadingMapModes

Details

When set to ON, android.statistics.lensShadingMap will be provided in the output result metadata.

ON is always supported on devices with the RAW capability.

static
Property Name Type Description Units Range Tags
android.statistics.info.availableFaceDetectModes byte x n [public as enumList] [legacy]
List of enums from android.statistics.faceDetectMode

List of face detection modes for android.statistics.faceDetectMode that are supported by this camera device.

Any value listed in android.statistics.faceDetectMode

Details

OFF is always supported.

android.statistics.info.histogramBucketCount int32 [system]

Number of histogram buckets supported

>= 64

android.statistics.info.maxFaceCount int32 [public] [legacy]

The maximum number of simultaneously detectable faces.

0 for cameras without available face detection; otherwise: >=4 for LIMITED or FULL hwlevel devices or >0 for LEGACY devices.

android.statistics.info.maxHistogramCount int32 [system]

Maximum value possible for a histogram bucket

android.statistics.info.maxSharpnessMapValue int32 [system]

Maximum value possible for a sharpness map region.

android.statistics.info.sharpnessMapSize int32 x 2 [system as size]
width x height

Dimensions of the sharpness map

Must be at least 32 x 32

android.statistics.info.availableHotPixelMapModes byte x n [public as boolean]
list of enums

List of hot pixel map output modes for android.statistics.hotPixelMapMode that are supported by this camera device.

Any value listed in android.statistics.hotPixelMapMode

Details

If no hotpixel map output is available for this camera device, this will contain only false.

ON is always supported on devices with the RAW capability.

android.statistics.info.availableLensShadingMapModes byte x n [public as enumList]
list of enums

List of lens shading map output modes for android.statistics.lensShadingMapMode that are supported by this camera device.

Any value listed in android.statistics.lensShadingMapMode

Details

If no lens shading map output is available for this camera device, this key will contain only OFF.

ON is always supported on devices with the RAW capability. LEGACY mode devices will always only support OFF.

dynamic
Property Name Type Description Units Range Tags
android.statistics.faceDetectMode byte [public] [legacy]
  • OFF

    Do not include face detection statistics in capture results.

  • SIMPLE [optional]

    Return face rectangle and confidence values only.

  • FULL [optional]

    Return all face metadata.

    In this mode, face rectangles, scores, landmarks, and face IDs are all valid.

Operating mode for the face detector unit.

android.statistics.info.availableFaceDetectModes

Details

Whether face detection is enabled, and whether it should output just the basic fields or the full set of fields.

HAL Implementation Details

SIMPLE mode must fill in android.statistics.faceRectangles and android.statistics.faceScores. FULL mode must also fill in android.statistics.faceIds, and android.statistics.faceLandmarks.

android.statistics.faceIds int32 x n [hidden] [legacy]

List of unique IDs for detected faces.

Details

Each detected face is given a unique ID that is valid for as long as the face is visible to the camera device. A face that leaves the field of view and later returns may be assigned a new ID.

Only available if android.statistics.faceDetectMode == FULL

android.statistics.faceLandmarks int32 x n x 6 [hidden] [legacy]
(leftEyeX, leftEyeY, rightEyeX, rightEyeY, mouthX, mouthY)

List of landmarks for detected faces.

Details

The coordinate system is that of android.sensor.info.activeArraySize, with (0, 0) being the top-left pixel of the active array.

Only available if android.statistics.faceDetectMode == FULL

android.statistics.faceRectangles int32 x n x 4 [hidden as rectangle] [legacy]
(xmin, ymin, xmax, ymax). (0,0) is top-left of active pixel area

List of the bounding rectangles for detected faces.

Details

The coordinate system is that of android.sensor.info.activeArraySize, with (0, 0) being the top-left pixel of the active array.

Only available if android.statistics.faceDetectMode != OFF

android.statistics.faceScores byte x n [hidden] [legacy]

List of the face confidence scores for detected faces

1-100

Details

Only available if android.statistics.faceDetectMode != OFF.

HAL Implementation Details

The value should be meaningful (for example, setting 100 at all times is illegal).

android.statistics.faces int32 x n [public as face] [synthetic] [legacy]

List of the faces detected through camera face detection in this capture.

Details

Only available if android.statistics.faceDetectMode != OFF.

android.statistics.histogram int32 x n x 3 [system]
count of pixels for each color channel that fall into each histogram bucket, scaled to be between 0 and maxHistogramCount

A 3-channel histogram based on the raw sensor data

Details

The k'th bucket (0-based) covers the input range (with w = android.sensor.info.whiteLevel) of [ k * w/N, (k + 1) * w / N ). If only a monochrome sharpness map is supported, all channels should have the same data

android.statistics.histogramMode byte [system as boolean]
  • OFF
  • ON

Operating mode for histogram generation

android.statistics.sharpnessMap int32 x n x m x 3 [system]
estimated sharpness for each region of the input image. Normalized to be between 0 and maxSharpnessMapValue. Higher values mean sharper (better focused)

A 3-channel sharpness map, based on the raw sensor data

Details

If only a monochrome sharpness map is supported, all channels should have the same data

android.statistics.sharpnessMapMode byte [system as boolean]
  • OFF
  • ON

Operating mode for sharpness map generation

android.statistics.lensShadingCorrectionMap byte [public as lensShadingMap] [full]

The shading map is a low-resolution floating-point map that lists the coefficients used to correct for vignetting, for each Bayer color channel.

Each gain factor is >= 1

Details

The least shaded section of the image should have a gain factor of 1; all other sections should have gains above 1.

When android.colorCorrection.mode = TRANSFORM_MATRIX, the map must take into account the colorCorrection settings.

The shading map is for the entire active pixel array, and is not affected by the crop region specified in the request. Each shading map entry is the value of the shading compensation map over a specific pixel on the sensor. Specifically, with a (N x M) resolution shading map, and an active pixel array size (W x H), shading map entry (x,y) ϵ (0 ... N-1, 0 ... M-1) is the value of the shading map at pixel ( ((W-1)/(N-1)) * x, ((H-1)/(M-1)) * y) for the four color channels. The map is assumed to be bilinearly interpolated between the sample points.

The channel order is [R, Geven, Godd, B], where Geven is the green channel for the even rows of a Bayer pattern, and Godd is the odd rows. The shading map is stored in a fully interleaved format.

The shading map should have on the order of 30-40 rows and columns, and must be smaller than 64x64.

As an example, given a very small map defined as:

width,height = [ 4, 3 ]
values =
[ 1.3, 1.2, 1.15, 1.2,  1.2, 1.2, 1.15, 1.2,
    1.1, 1.2, 1.2, 1.2,  1.3, 1.2, 1.3, 1.3,
  1.2, 1.2, 1.25, 1.1,  1.1, 1.1, 1.1, 1.0,
    1.0, 1.0, 1.0, 1.0,  1.2, 1.3, 1.25, 1.2,
  1.3, 1.2, 1.2, 1.3,   1.2, 1.15, 1.1, 1.2,
    1.2, 1.1, 1.0, 1.2,  1.3, 1.15, 1.2, 1.3 ]

The low-resolution scaling map images for each channel are (displayed using nearest-neighbor interpolation):

Red lens shading map Green (even rows) lens shading map Green (odd rows) lens shading map Blue lens shading map

As a visualization only, inverting the full-color map to recover an image of a gray wall (using bicubic interpolation for visual quality) as captured by the sensor gives:

Image of a uniform white wall (inverse shading map)

android.statistics.lensShadingMap float x 4 x n x m [hidden] [full]
2D array of float gain factors per channel to correct lens shading

The shading map is a low-resolution floating-point map that lists the coefficients used to correct for vignetting, for each Bayer color channel of RAW image data.

Each gain factor is >= 1

Details

The least shaded section of the image should have a gain factor of 1; all other sections should have gains above 1.

When android.colorCorrection.mode = TRANSFORM_MATRIX, the map must take into account the colorCorrection settings.

The shading map is for the entire active pixel array, and is not affected by the crop region specified in the request. Each shading map entry is the value of the shading compensation map over a specific pixel on the sensor. Specifically, with a (N x M) resolution shading map, and an active pixel array size (W x H), shading map entry (x,y) ϵ (0 ... N-1, 0 ... M-1) is the value of the shading map at pixel ( ((W-1)/(N-1)) * x, ((H-1)/(M-1)) * y) for the four color channels. The map is assumed to be bilinearly interpolated between the sample points.

The channel order is [R, Geven, Godd, B], where Geven is the green channel for the even rows of a Bayer pattern, and Godd is the odd rows. The shading map is stored in a fully interleaved format, and its size is provided in the camera static metadata by android.lens.info.shadingMapSize.

The shading map should have on the order of 30-40 rows and columns, and must be smaller than 64x64.

As an example, given a very small map defined as:

android.lens.info.shadingMapSize = [ 4, 3 ]
android.statistics.lensShadingMap =
[ 1.3, 1.2, 1.15, 1.2,  1.2, 1.2, 1.15, 1.2,
    1.1, 1.2, 1.2, 1.2,  1.3, 1.2, 1.3, 1.3,
  1.2, 1.2, 1.25, 1.1,  1.1, 1.1, 1.1, 1.0,
    1.0, 1.0, 1.0, 1.0,  1.2, 1.3, 1.25, 1.2,
  1.3, 1.2, 1.2, 1.3,   1.2, 1.15, 1.1, 1.2,
    1.2, 1.1, 1.0, 1.2,  1.3, 1.15, 1.2, 1.3 ]

The low-resolution scaling map images for each channel are (displayed using nearest-neighbor interpolation):

Red lens shading map Green (even rows) lens shading map Green (odd rows) lens shading map Blue lens shading map

As a visualization only, inverting the full-color map to recover an image of a gray wall (using bicubic interpolation for visual quality) as captured by the sensor gives:

Image of a uniform white wall (inverse shading map)

Note that the RAW image data might be subject to lens shading correction not reported on this map. Query android.sensor.info.lensShadingApplied to see if RAW image data has subject to lens shading correction. If android.sensor.info.lensShadingApplied is TRUE, the RAW image data is subject to partial or full lens shading correction. In the case full lens shading correction is applied to RAW images, the gain factor map reported in this key will contain all 1.0 gains. In other words, the map reported in this key is the remaining lens shading that needs to be applied on the RAW image to get images without lens shading artifacts. See android.request.maxNumOutputRaw for a list of RAW image formats.

HAL Implementation Details

The lens shading map calculation may depend on exposure and white balance statistics. When AE and AWB are in AUTO modes (android.control.aeMode != OFF and android.control.awbMode != OFF), the HAL may have all the information it need to generate most accurate lens shading map. When AE or AWB are in manual mode (android.control.aeMode == OFF or android.control.awbMode == OFF), the shading map may be adversely impacted by manual exposure or white balance parameters. To avoid generating unreliable shading map data, the HAL may choose to lock the shading map with the latest known good map generated when the AE and AWB are in AUTO modes.

android.statistics.predictedColorGains float x 4 [hidden] [deprecated]
A 1D array of floats for 4 color channel gains

The best-fit color channel gains calculated by the camera device's statistics units for the current output frame.

Deprecated. Do not use.

Details

This may be different than the gains used for this frame, since statistics processing on data from a new frame typically completes after the transform has already been applied to that frame.

The 4 channel gains are defined in Bayer domain, see android.colorCorrection.gains for details.

This value should always be calculated by the auto-white balance (AWB) block, regardless of the android.control.* current values.

android.statistics.predictedColorTransform rational x 3 x 3 [hidden] [deprecated]
3x3 rational matrix in row-major order

The best-fit color transform matrix estimate calculated by the camera device's statistics units for the current output frame.

Deprecated. Do not use.

Details

The camera device will provide the estimate from its statistics unit on the white balance transforms to use for the next frame. These are the values the camera device believes are the best fit for the current output frame. This may be different than the transform used for this frame, since statistics processing on data from a new frame typically completes after the transform has already been applied to that frame.

These estimates must be provided for all frames, even if capture settings and color transforms are set by the application.

This value should always be calculated by the auto-white balance (AWB) block, regardless of the android.control.* current values.

android.statistics.sceneFlicker byte [public] [full]
  • NONE

    The camera device does not detect any flickering illumination in the current scene.

  • 50HZ

    The camera device detects illumination flickering at 50Hz in the current scene.

  • 60HZ

    The camera device detects illumination flickering at 60Hz in the current scene.

The camera device estimated scene illumination lighting frequency.

Details

Many light sources, such as most fluorescent lights, flicker at a rate that depends on the local utility power standards. This flicker must be accounted for by auto-exposure routines to avoid artifacts in captured images. The camera device uses this entry to tell the application what the scene illuminant frequency is.

When manual exposure control is enabled (android.control.aeMode == OFF or android.control.mode == OFF), the android.control.aeAntibandingMode doesn't perform antibanding, and the application can ensure it selects exposure times that do not cause banding issues by looking into this metadata field. See android.control.aeAntibandingMode for more details.

Reports NONE if there doesn't appear to be flickering illumination.

android.statistics.hotPixelMapMode byte [public as boolean]
  • OFF

    Hot pixel map production is disabled.

  • ON

    Hot pixel map production is enabled.

Operating mode for hot pixel map generation.

android.statistics.info.availableHotPixelMapModes

Details

If set to true, a hot pixel map is returned in android.statistics.hotPixelMap. If set to false, no hot pixel map will be returned.

android.statistics.hotPixelMap int32 x 2 x n [public as point]
list of coordinates based on android.sensor.pixelArraySize

List of (x, y) coordinates of hot/defective pixels on the sensor.

n <= number of pixels on the sensor. The (x, y) coordinates must be bounded by android.sensor.info.pixelArraySize.

Details

A coordinate (x, y) must lie between (0, 0), and (width - 1, height - 1) (inclusive), which are the top-left and bottom-right of the pixel array, respectively. The width and height dimensions are given in android.sensor.info.pixelArraySize. This may include hot pixels that lie outside of the active array bounds given by android.sensor.info.activeArraySize.

HAL Implementation Details

A hotpixel map contains the coordinates of pixels on the camera sensor that do report valid values (usually due to defects in the camera sensor). This includes pixels that are stuck at certain values, or have a response that does not accuractly encode the incoming light from the scene.

To avoid performance issues, there should be significantly fewer hot pixels than actual pixels on the camera sensor.

android.statistics.lensShadingMapMode byte [public] [full]
  • OFF

    Do not include a lens shading map in the capture result.

  • ON

    Include a lens shading map in the capture result.

Whether the camera device will output the lens shading map in output result metadata.

android.statistics.info.availableLensShadingMapModes

Details

When set to ON, android.statistics.lensShadingMap will be provided in the output result metadata.

ON is always supported on devices with the RAW capability.

tonemap
controls
Property Name Type Description Units Range Tags
android.tonemap.curveBlue float x n x 2 [hidden] [full]
1D array of float pairs (P_IN, P_OUT). The maximum number of pairs is specified by android.tonemap.maxCurvePoints.

Tonemapping / contrast / gamma curve for the blue channel, to use when android.tonemap.mode is CONTRAST_CURVE.

Details

See android.tonemap.curveRed for more details.

android.tonemap.curveGreen float x n x 2 [hidden] [full]
1D array of float pairs (P_IN, P_OUT). The maximum number of pairs is specified by android.tonemap.maxCurvePoints.

Tonemapping / contrast / gamma curve for the green channel, to use when android.tonemap.mode is CONTRAST_CURVE.

Details

See android.tonemap.curveRed for more details.

android.tonemap.curveRed float x n x 2 [hidden] [full]
1D array of float pairs (P_IN, P_OUT). The maximum number of pairs is specified by android.tonemap.maxCurvePoints.

Tonemapping / contrast / gamma curve for the red channel, to use when android.tonemap.mode is CONTRAST_CURVE.

0-1 on both input and output coordinates, normalized as a floating-point value such that 0 == black and 1 == white.

Details

Each channel's curve is defined by an array of control points:

android.tonemap.curveRed =
  [ P0in, P0out, P1in, P1out, P2in, P2out, P3in, P3out, ..., PNin, PNout ]
2 <= N <= android.tonemap.maxCurvePoints

These are sorted in order of increasing Pin; it is required that input values 0.0 and 1.0 are included in the list to define a complete mapping. For input values between control points, the camera device must linearly interpolate between the control points.

Each curve can have an independent number of points, and the number of points can be less than max (that is, the request doesn't have to always provide a curve with number of points equivalent to android.tonemap.maxCurvePoints).

A few examples, and their corresponding graphical mappings; these only specify the red channel and the precision is limited to 4 digits, for conciseness.

Linear mapping:

android.tonemap.curveRed = [ 0, 0, 1.0, 1.0 ]

Linear mapping curve

Invert mapping:

android.tonemap.curveRed = [ 0, 1.0, 1.0, 0 ]

Inverting mapping curve

Gamma 1/2.2 mapping, with 16 control points:

android.tonemap.curveRed = [
  0.0000, 0.0000, 0.0667, 0.2920, 0.1333, 0.4002, 0.2000, 0.4812,
  0.2667, 0.5484, 0.3333, 0.6069, 0.4000, 0.6594, 0.4667, 0.7072,
  0.5333, 0.7515, 0.6000, 0.7928, 0.6667, 0.8317, 0.7333, 0.8685,
  0.8000, 0.9035, 0.8667, 0.9370, 0.9333, 0.9691, 1.0000, 1.0000 ]

Gamma = 1/2.2 tonemapping curve

Standard sRGB gamma mapping, per IEC 61966-2-1:1999, with 16 control points:

android.tonemap.curveRed = [
  0.0000, 0.0000, 0.0667, 0.2864, 0.1333, 0.4007, 0.2000, 0.4845,
  0.2667, 0.5532, 0.3333, 0.6125, 0.4000, 0.6652, 0.4667, 0.7130,
  0.5333, 0.7569, 0.6000, 0.7977, 0.6667, 0.8360, 0.7333, 0.8721,
  0.8000, 0.9063, 0.8667, 0.9389, 0.9333, 0.9701, 1.0000, 1.0000 ]

sRGB tonemapping curve

HAL Implementation Details

For good quality of mapping, at least 128 control points are preferred.

A typical use case of this would be a gamma-1/2.2 curve, with as many control points used as are available.

android.tonemap.curve float [public as tonemapCurve] [synthetic] [full]

Tonemapping / contrast / gamma curve to use when android.tonemap.mode is CONTRAST_CURVE.

Details

The tonemapCurve consist of three curves for each of red, green, and blue channels respectively. The following example uses the red channel as an example. The same logic applies to green and blue channel. Each channel's curve is defined by an array of control points:

curveRed =
  [ P0(in, out), P1(in, out), P2(in, out), P3(in, out), ..., PN(in, out) ]
2 <= N <= android.tonemap.maxCurvePoints

These are sorted in order of increasing Pin; it is always guaranteed that input values 0.0 and 1.0 are included in the list to define a complete mapping. For input values between control points, the camera device must linearly interpolate between the control points.

Each curve can have an independent number of points, and the number of points can be less than max (that is, the request doesn't have to always provide a curve with number of points equivalent to android.tonemap.maxCurvePoints).

A few examples, and their corresponding graphical mappings; these only specify the red channel and the precision is limited to 4 digits, for conciseness.

Linear mapping:

curveRed = [ (0, 0), (1.0, 1.0) ]

Linear mapping curve

Invert mapping:

curveRed = [ (0, 1.0), (1.0, 0) ]

Inverting mapping curve

Gamma 1/2.2 mapping, with 16 control points:

curveRed = [
  (0.0000, 0.0000), (0.0667, 0.2920), (0.1333, 0.4002), (0.2000, 0.4812),
  (0.2667, 0.5484), (0.3333, 0.6069), (0.4000, 0.6594), (0.4667, 0.7072),
  (0.5333, 0.7515), (0.6000, 0.7928), (0.6667, 0.8317), (0.7333, 0.8685),
  (0.8000, 0.9035), (0.8667, 0.9370), (0.9333, 0.9691), (1.0000, 1.0000) ]

Gamma = 1/2.2 tonemapping curve

Standard sRGB gamma mapping, per IEC 61966-2-1:1999, with 16 control points:

curveRed = [
  (0.0000, 0.0000), (0.0667, 0.2864), (0.1333, 0.4007), (0.2000, 0.4845),
  (0.2667, 0.5532), (0.3333, 0.6125), (0.4000, 0.6652), (0.4667, 0.7130),
  (0.5333, 0.7569), (0.6000, 0.7977), (0.6667, 0.8360), (0.7333, 0.8721),
  (0.8000, 0.9063), (0.8667, 0.9389), (0.9333, 0.9701), (1.0000, 1.0000) ]

sRGB tonemapping curve

HAL Implementation Details

This entry is created by the framework from the curveRed, curveGreen and curveBlue entries.

android.tonemap.mode byte [public] [full]
  • CONTRAST_CURVE

    Use the tone mapping curve specified in the android.tonemap.curve* entries.

    All color enhancement and tonemapping must be disabled, except for applying the tonemapping curve specified by android.tonemap.curve.

    Must not slow down frame rate relative to raw sensor output.

  • FAST

    Advanced gamma mapping and color enhancement may be applied, without reducing frame rate compared to raw sensor output.

  • HIGH_QUALITY

    High-quality gamma mapping and color enhancement will be applied, at the cost of possibly reduced frame rate compared to raw sensor output.

  • GAMMA_VALUE

    Use the gamma value specified in android.tonemap.gamma to peform tonemapping.

    All color enhancement and tonemapping must be disabled, except for applying the tonemapping curve specified by android.tonemap.gamma.

    Must not slow down frame rate relative to raw sensor output.

  • PRESET_CURVE

    Use the preset tonemapping curve specified in android.tonemap.presetCurve to peform tonemapping.

    All color enhancement and tonemapping must be disabled, except for applying the tonemapping curve specified by android.tonemap.presetCurve.

    Must not slow down frame rate relative to raw sensor output.

High-level global contrast/gamma/tonemapping control.

android.tonemap.availableToneMapModes

Details

When switching to an application-defined contrast curve by setting android.tonemap.mode to CONTRAST_CURVE, the curve is defined per-channel with a set of (in, out) points that specify the mapping from input high-bit-depth pixel value to the output low-bit-depth value. Since the actual pixel ranges of both input and output may change depending on the camera pipeline, the values are specified by normalized floating-point numbers.

More-complex color mapping operations such as 3D color look-up tables, selective chroma enhancement, or other non-linear color transforms will be disabled when android.tonemap.mode is CONTRAST_CURVE.

When using either FAST or HIGH_QUALITY, the camera device will emit its own tonemap curve in android.tonemap.curve. These values are always available, and as close as possible to the actually used nonlinear/nonglobal transforms.

If a request is sent with CONTRAST_CURVE with the camera device's provided curve in FAST or HIGH_QUALITY, the image's tonemap will be roughly the same.

android.tonemap.gamma float [public]

Tonemapping curve to use when android.tonemap.mode is GAMMA_VALUE

Details

The tonemap curve will be defined the following formula: * OUT = pow(IN, 1.0 / gamma) where IN and OUT is the input pixel value scaled to range [0.0, 1.0], pow is the power function and gamma is the gamma value specified by this key.

The same curve will be applied to all color channels. The camera device may clip the input gamma value to its supported range. The actual applied value will be returned in capture result.

The valid range of gamma value varies on different devices, but values within [1.0, 5.0] are guaranteed not to be clipped.

android.tonemap.presetCurve byte [public]
  • SRGB

    Tonemapping curve is defined by sRGB

  • REC709

    Tonemapping curve is defined by ITU-R BT.709

Tonemapping curve to use when android.tonemap.mode is PRESET_CURVE

Details

The tonemap curve will be defined by specified standard.

sRGB (approximated by 16 control points):

sRGB tonemapping curve

Rec. 709 (approximated by 16 control points):

Rec. 709 tonemapping curve

Note that above figures show a 16 control points approximation of preset curves. Camera devices may apply a different approximation to the curve.

static
Property Name Type Description Units Range Tags
android.tonemap.maxCurvePoints int32 [public] [full]

Maximum number of supported points in the tonemap curve that can be used for android.tonemap.curve.

Details

If the actual number of points provided by the application (in android.tonemap.curve*) is less than this maximum, the camera device will resample the curve to its internal representation, using linear interpolation.

The output curves in the result metadata may have a different number of points than the input curves, and will represent the actual hardware curves used as closely as possible when linearly interpolated.

HAL Implementation Details

This value must be at least 64. This should be at least 128.

android.tonemap.availableToneMapModes byte x n [public as enumList] [full]
list of enums

List of tonemapping modes for android.tonemap.mode that are supported by this camera device.

Any value listed in android.tonemap.mode

Details

Camera devices that support the MANUAL_POST_PROCESSING capability will always contain at least one of below mode combinations:

  • CONTRAST_CURVE, FAST and HIGH_QUALITY
  • GAMMA_VALUE, PRESET_CURVE, FAST and HIGH_QUALITY

This includes all FULL level devices.

HAL Implementation Details

HAL must support both FAST and HIGH_QUALITY if automatic tonemap control is available on the camera device, but the underlying implementation can be the same for both modes. That is, if the highest quality implementation on the camera device does not slow down capture rate, then FAST and HIGH_QUALITY will generate the same output.

dynamic
Property Name Type Description Units Range Tags
android.tonemap.curveBlue float x n x 2 [hidden] [full]
1D array of float pairs (P_IN, P_OUT). The maximum number of pairs is specified by android.tonemap.maxCurvePoints.

Tonemapping / contrast / gamma curve for the blue channel, to use when android.tonemap.mode is CONTRAST_CURVE.

Details

See android.tonemap.curveRed for more details.

android.tonemap.curveGreen float x n x 2 [hidden] [full]
1D array of float pairs (P_IN, P_OUT). The maximum number of pairs is specified by android.tonemap.maxCurvePoints.

Tonemapping / contrast / gamma curve for the green channel, to use when android.tonemap.mode is CONTRAST_CURVE.

Details

See android.tonemap.curveRed for more details.

android.tonemap.curveRed float x n x 2 [hidden] [full]
1D array of float pairs (P_IN, P_OUT). The maximum number of pairs is specified by android.tonemap.maxCurvePoints.

Tonemapping / contrast / gamma curve for the red channel, to use when android.tonemap.mode is CONTRAST_CURVE.

0-1 on both input and output coordinates, normalized as a floating-point value such that 0 == black and 1 == white.

Details

Each channel's curve is defined by an array of control points:

android.tonemap.curveRed =
  [ P0in, P0out, P1in, P1out, P2in, P2out, P3in, P3out, ..., PNin, PNout ]
2 <= N <= android.tonemap.maxCurvePoints

These are sorted in order of increasing Pin; it is required that input values 0.0 and 1.0 are included in the list to define a complete mapping. For input values between control points, the camera device must linearly interpolate between the control points.

Each curve can have an independent number of points, and the number of points can be less than max (that is, the request doesn't have to always provide a curve with number of points equivalent to android.tonemap.maxCurvePoints).

A few examples, and their corresponding graphical mappings; these only specify the red channel and the precision is limited to 4 digits, for conciseness.

Linear mapping:

android.tonemap.curveRed = [ 0, 0, 1.0, 1.0 ]

Linear mapping curve

Invert mapping:

android.tonemap.curveRed = [ 0, 1.0, 1.0, 0 ]

Inverting mapping curve

Gamma 1/2.2 mapping, with 16 control points:

android.tonemap.curveRed = [
  0.0000, 0.0000, 0.0667, 0.2920, 0.1333, 0.4002, 0.2000, 0.4812,
  0.2667, 0.5484, 0.3333, 0.6069, 0.4000, 0.6594, 0.4667, 0.7072,
  0.5333, 0.7515, 0.6000, 0.7928, 0.6667, 0.8317, 0.7333, 0.8685,
  0.8000, 0.9035, 0.8667, 0.9370, 0.9333, 0.9691, 1.0000, 1.0000 ]

Gamma = 1/2.2 tonemapping curve

Standard sRGB gamma mapping, per IEC 61966-2-1:1999, with 16 control points:

android.tonemap.curveRed = [
  0.0000, 0.0000, 0.0667, 0.2864, 0.1333, 0.4007, 0.2000, 0.4845,
  0.2667, 0.5532, 0.3333, 0.6125, 0.4000, 0.6652, 0.4667, 0.7130,
  0.5333, 0.7569, 0.6000, 0.7977, 0.6667, 0.8360, 0.7333, 0.8721,
  0.8000, 0.9063, 0.8667, 0.9389, 0.9333, 0.9701, 1.0000, 1.0000 ]

sRGB tonemapping curve

HAL Implementation Details

For good quality of mapping, at least 128 control points are preferred.

A typical use case of this would be a gamma-1/2.2 curve, with as many control points used as are available.

android.tonemap.curve float [public as tonemapCurve] [synthetic] [full]

Tonemapping / contrast / gamma curve to use when android.tonemap.mode is CONTRAST_CURVE.

Details

The tonemapCurve consist of three curves for each of red, green, and blue channels respectively. The following example uses the red channel as an example. The same logic applies to green and blue channel. Each channel's curve is defined by an array of control points:

curveRed =
  [ P0(in, out), P1(in, out), P2(in, out), P3(in, out), ..., PN(in, out) ]
2 <= N <= android.tonemap.maxCurvePoints

These are sorted in order of increasing Pin; it is always guaranteed that input values 0.0 and 1.0 are included in the list to define a complete mapping. For input values between control points, the camera device must linearly interpolate between the control points.

Each curve can have an independent number of points, and the number of points can be less than max (that is, the request doesn't have to always provide a curve with number of points equivalent to android.tonemap.maxCurvePoints).

A few examples, and their corresponding graphical mappings; these only specify the red channel and the precision is limited to 4 digits, for conciseness.

Linear mapping:

curveRed = [ (0, 0), (1.0, 1.0) ]

Linear mapping curve

Invert mapping:

curveRed = [ (0, 1.0), (1.0, 0) ]

Inverting mapping curve

Gamma 1/2.2 mapping, with 16 control points:

curveRed = [
  (0.0000, 0.0000), (0.0667, 0.2920), (0.1333, 0.4002), (0.2000, 0.4812),
  (0.2667, 0.5484), (0.3333, 0.6069), (0.4000, 0.6594), (0.4667, 0.7072),
  (0.5333, 0.