[Feature Request] Add support to vs-ddcolor

[Selur]

I noted that by enabling FP16 computation the conversion of color space applied is RGBH, while when FP16 is uncecked is applied RGBS.

Yes, that is the difference between FP16 and FP32. One works with 16bit precision (RGBH) and the other works with 32bit precision (RGBS).

Only RGBH and RGBS formats are supported. RGBH uses the bfloat16 data type for inference while RGBS uses the float32 data type.

source: https://github.com/HolyWu/vs-ddcolor/blo...C54-L28C99

Do you know if there is other significant difference is using the 2 color spaces in terms of final image quality ?

Usually, the difference between using 16bit and 32bit is minimal at best.

Cu Selur