[FFmpeg-devel] A potential problem...

Art Clarke aclarke
Tue Mar 24 22:51:00 CET 2009

So I've confirmed it's a bug in tip of tree swscale when converting from
PIX_FMT_YUV420P to PIX_FMT_ARGB and back.

On Mon, Mar 23, 2009 at 6:34 PM, compn <tempn at twmi.rr.com> wrote:

> bugreports > roundup


> >I'm not pointing the finger at FFMPEG yet -- to do I know I need a
> >reproducible test case written in libav that shows the problem (which
> we're

now I am :)

> you dont need that (i dont think?). just a simple ffmpeg line w/
> small sample input video file to compare svn ffmpeg to old ffmpeg and
> thats it.

Actually I'm not sure how to use ffmpeg command line to reproduce, but I did
attach a very simple repro case using libswscale in output_example.c that
quickly shows the problem.  It converts the output image from YUV420P to
ARGB and back, and voila, the resulting image looks wrong.

Use RGB24 instead of ARGB as the temporary picture and the conversion works
just fine.

I haven't tried to patch it yet because there's been A LOT of swscale
changes over the last few days, including assembly, and I'm assuming someone
is better equipped to deal with it.  Let me know if that's not the case, and
I'll dig in.

- Art

xu?ggle (z?' gl) v. To freely encode, decode, and experience audio and

Use Xuggle to get the power of FFMPEG in Java.

More information about the ffmpeg-devel mailing list