The shading map is a low-resolution floating-point map that lists the coefficients used to correct for vignetting, for each Bayer color channel.

The least shaded section of the image should have a gain factor of 1; all other sections should have gains above 1.

When CaptureRequest.ColorCorrectionMode = 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:

The low-resolution scaling map images for each channel are (displayed using nearest-neighbor interpolation):

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:

Range of valid values:

Each gain factor is >= 1

Optional - This value may be null on some devices.

Full capability - Present on all camera devices that report being CameraMetadata.INFO_SUPPORTED_HARDWARE_LEVEL_FULL devices in the CameraCharacteristics.InfoSupportedHardwareLevel key

[Android Documentation]