V4l2 returning 14-bit raw instead of 12-bit. What is going on?

[JDSchroeder]

NVIDIA’s raw memory for RAW12 uses 14-bits. The raw image modes either replicate MSBs (for T_R16 16-bit bpp widths) or truncate LSBs (for T_L8 8-bit bpp widths), as needed to fit pixels into the appropriate format.

See this thread for details:

Raw Bayer GRBG 12-bit Image Data from CSI-2

I found this picture in the TRM (Figure 198, section 27.10.6 RAW Memory Formats): [image] It would make more sense to me if all of the RAW formats were shifted to have the valid data put in the MSBit. Why 14-bits? Why the “implied decimal point”?

So for RAW12 you get bits 15 and 14 zero’d and bits 1 and 0 are replications of the two MSBits (D11 and D10).