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

[Marton Balint]

On Tue, 8 Jun 2021, Valerii Zapodovnikov wrote:

> Can you merge my mxfdec patch? Thank you.

I'd give at least a week for the mxf maintainer to respond before 
pushing.

> Listen, it is commonly known that ffplay is broken, Carl agrees. MPV is not
> broken and is perfect according to reference calculator here.
> https://res18h39.netlify.app/color
>
> I did send you a sample in my "v1" patch (this is not v2, since v1 had
> wrong title, so v2 would be wrong thing to do, I just superseeded patch on
> patchwork). But for your comfort here it is again:

Patchwork statuses are rarely monitored, so IMHO it is better to use v2 in 
the mailing list even for cases like this.

> https://trac.ffmpeg.org/attachment/ticket/9167/2.mp4 (also read #9167).
>
> Just *use* the sample above and ffplay it without any flags and ffplay
> with -vf scale=in_color_matrix=bt601,format=gbrp.

Both attempts gets me 238,77,44. If you get something else, then I guess 
it is SDL version / renderer / OS / HW dependant? With MPV, it is 
237,77,44 for me.

> I do not really care about ffplay,

Ok.

> I care about Chromium (since I work on
> it) and now VLC. You can carefully read those two issues that use the
> sample above too.
>
> https://code.videolan.org/videolan/vlc/-/issues/25651
>
> https://code.videolan.org/videolan/vlc/-/issues/25676
>
> As you can see from those issues I know a lot about YCbCr and read the
> first documents from back when ITU-R was CCI(R). I even know that BT.709
> was first defined in CCIR Rec. XA/11 1986-90f, a.k.a. IWP 11/6-2108
> (Canada). That is not something you can fimd on the Internet, only in ITU-R
> archives (wikipedia is me, I wrote it there). So, please, stop with this.

I do respect your knowledge regarding colorspaces, but I cannot reproduce 
the issue in 8862, therefore the FIXME makes no sense to me. So applied 
the patch without it.

Regards,
Marton