Section 1 - The Depth Field

When filling in the depth field directly, use the following guide:

• If the image has an alpha channel, always use 32.

• If the image is grayscale, use a grayscale depth:

  • 40 for 8-bit or deeper samples

  • 36 for 4-bit samples

  • 34 for 2-bit samples

  • 33 for 1-bit samples

• If the image is indexed color (ie, each sample is interpreted through a color look-up table), use the appropriate color depth; In these cases, the color table must be attached to the image description. You can do this by calling SetImageDescriptionCTable.

  • 8 for 256 color

  • 4 for 16 color

  • 2 for 4 color

  • 1 for Monochrome

• If the image is a 16-bit RGB pixel format, use 16.

Otherwise use 24 if the image doesn't match any of the above criteria. This is the generic "color with no alpha channel" depth.

For YUV pixel formats without alpha, use 24. Although YUV9 is 9 bits per pixel, YUV 4:2:0 is 12 bits per pixel, and YUV 4:2:2 is 16 bits per pixel, you should still use 24.

Follow the above guidelines even for pixel formats with samples deeper than 8-bit. For example, 48-bit RGB should use depth 24 and 64-bit ARGB should use depth 32.

QTGetPixelFormatDepthForImageDescription (available in QuickTime 6.0 and greater) returns the appropriate depth code for any registered pixel format.

Section 2 - Where should information about the true depth get stored?

When the image data in question is an uncompressed pixel array of a big-endian pixel format that QuickDraw supports (specifically, 1, 2, 4, 8, 16, 24, 32, 33, 34, 36, or 40), the cType field of the image description should be kRawCodecType and the depth field (as discussed above) should be the pixel format. For example, for 8-bit grayscale, the cType field should contain kRawCodecType and the depth field should contain 40.

When the image data in question is an uncompressed pixel array of a non-QuickDraw pixel format, the cType field of the image description should be the pixel format code. For example, k2vuyPixelFormat ('2vuy'). In this case the depth field (as discussed above) should contain 24. For 16-bit grayscale, the cType field should contain k16GrayCodecType and the depth field should contain 40.

When the image data in question is compressed, but has a special native pixel format that it can be decompressed directly into, the pixel format should be stored as a big-endian OSType in an image description extension of type kImageDescriptionColorSpace ('cspc'). The value of the cType field should reflect the way the image data was compressed.

For example, a PNG file containing a 48-bit-per-pixel RGB image would be described by an image description with a cType of kPNGCodecType ('png '), a big-endian kImageDescriptionColorSpace ('cspc') image description extension containing the OSType k48RGBCodecType ('b48r') and a depth field (as discussed above) containing the value 24.

If your application needs more detailed information about registered pixel formats, use ICMGetPixelFormatInfo.

References

• The Image Description Structure

• QTGetPixelFormatDepthForImageDescription - What's new in QuickTime 6

• SetImageDescriptionCTable

• ICMGetPixelFormatInfo

• Technical Note TN2057, 'Fill in the size field before calling ICMGetPixelFormatInfo'