[FFmpeg-devel] h264 decoding regressions

[Don Moir]

This is just a mention because I have not looked at it in detail yet.

Bunny plays fine when using SwsContext for converting from PIX_FMT_GBR24P to 
PIX_FMT_YUV420P

but doesn't appear to work when converting from PIX_FMT_GBR24P to an RGB 
format like PIX_FMT_BGRA.

I need to look at in more detail and will report it as bug if there is a 
problem, but does anyone have a heads up on this ?

This worked before when the colors were swaped, but I think now all I get is 
black when going to an RGB format. The reason I have to say 'I think' is 
because I use YUV for output when it applies, but I also use RGB formats for 
thumbnail generation. Thumbnail is all black. I will force display to goto 
RGB instead of YUV and see what happens there when I get a chance.
----- Original Message ----- 

From: "Carl Eugen Hoyos" <cehoyos at ag.or.at>
To: <ffmpeg-devel at ffmpeg.org>
Sent: Sunday, October 23, 2011 3:58 PM
Subject: Re: [FFmpeg-devel] h264 decoding regressions


> madshi <madshi <at> gmail.com> writes:
>
>> 2011/10/23 Carl Eugen Hoyos <cehoyos <at> ag.or.at>
>> > Given that a few days ago, I couldn't easily find a
>> > 444 sample, I wonder if "serious" is really correct
>>
>> Haha!
>>
>> Well, I have a couple of samples here I'm sometimes
>> testing against. And two of them which once worked
>> are now broken with the current sources, so I thought
>> that would classify as "serious" enough.
>
> To clarify: I think all regressions are serious.
>
>> > I believe this sample never played correctly, or am
>> > I wrong?
>>
>> It did work a while ago.
>
> I believe it (bunny) played incorrectly, but it does not matter: It plays 
> fine
> now.
>
> The second file (BBB...) does not crash (anymore?), does it also work for 
> you?
>
> Do you have a H.264 444 9-bit sample?
>
> Carl Eugen
>
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel