[FFmpeg-user] xyz to rgb conversion

[SK Cinema]

Carl Eugen Hoyos said:
>I may misunderstand but this sounds to me as if no software 
>would be able to recognize this file as XYZ or do I miss 
>something?

Yes, that's correct.

>As said elsewhere, please try to explain what is wrong with 
>the conversion by FFmpeg? I suspect (maybe I am wrong) that 
>no developer know why other coefficients may be needed.
>(In the sense of: Which usecases need other coefficients 
>and which ones.)

The current ffmpeg conversion would use matrix coefficients for XYZ to
Rec.709 conversion and assume specific gamma values for the source X'Y'Z'
image and the target Rec.709 RGB image. A possible alternative requirement
would be for converting to the DCI P3 colour space rather than Rec.709.
Also, one might wish to use different gamma values.

Also, XYZ is the color space with the widest gamut one would encounter.
There would likely be material that is out of gamut in Rec.709 and this can
produce nasty artifacts. Using a 3D LUT that is built to gently handle such
out of gamut issues would often be preferable to a matrix color space
conversion. 

For reference, here's the command line I'd like to use:
ffmpeg -i video-source.mxf -vf
scale=1920:808,lut3d=XYZ-sRGB.3dl,pad=width=1920:height=1080:x=0:y=136
-codec prores output.mov





--
View this message in context: http://ffmpeg-users.933282.n4.nabble.com/xyz-to-rgb-conversion-tp4661196p4661664.html
Sent from the FFmpeg-users mailing list archive at Nabble.com.