[FFmpeg-devel] [PATCH 2/2] fftools/ffprobe: 240M matrix is not the same as BT.601

[Valerii Zapodovnikov]

AVFrame->color_range? But even if I were using full range file (I do not)
that would force BT.601 full range matrix (that is JPEG matrix) and I have
the opposite here, BT.709 is forced.

I suppose the problem can be checked with simple printf
of frame->colorspace with that sample that uses BT.601. This one:

https://trac.ffmpeg.org/attachment/ticket/9167/2.mp4

It should be wrong at that point. Or use breakpoint in debugger. :)

Also, BTW, Nvidia drivers windows 10, what is matrix used for 4K YCbCr
formats by default? Is it already BT.709? Because xvYCC can be both BT.601
and BT.709 (but it is only limited range, even if full is set by mistake,
see latest "ANSI/CTA-861-H") and sYCC with selectable quantisation uses
BT.601 even on 4K! But those are so called digital formats, they should be
further forced.