[FFmpeg-devel] [BUG?] pixel format conversion not handled

[Kostya]

On Fri, Jul 13, 2007 at 02:52:19PM +0200, Michael Niedermayer wrote:
> Hi
> 
> On Fri, Jul 13, 2007 at 07:06:35AM +0300, Kostya wrote:
> > On Thu, Jul 12, 2007 at 10:32:25PM +0200, Michael Niedermayer wrote:
> > > Hi
> > > 
> > > On Thu, Jul 12, 2007 at 08:38:13PM +0300, Kostya wrote:
> > > > I'm trying to test my decoder (svn://svn.mplayerhq.hu/soc/rv40) and
> > > > for some reason ffmpeg says "pixel format conversion not supported"
> > > > and gives only black frame while ffplay plays it fine (that garbage
> > > > my decoder produces for now).
> > > > 
> > > > It happens because ffmpeg does not set the input image format.
> > > > Can this be caused by my code, demuxer or something is wrong in
> > > > ffmpeg.c?
> > > 
> > > ffplay_g Blade2_640_1Mbps.rm
> > > ...
> > > *** glibc detected *** ./ffplay_g: corrupted double-linked list: 0x0868f1b0 ***
> > > 
> > > maybe you should run this under valgrind ...
> > 
> > Found one place where memset goes out of bounds but this problem caused
> > by different means. 
> > 
> > It happens only on Sample.rmvb (http://samples.mplayerhq.hu/real/VC-RV40/)
> > and looks like it does so because of some additional unknown codec.
> > I must add this happens by passing icodec->pix_fmt=-1 to sws_getContext()
> > even before codec init. 
> 
> try -analyzeduration 9M

It works, even -analyzeduration 6M is enough
 
> [...]
> -- 
> Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
> 
> It is dangerous to be right in matters on which the established authorities
> are wrong. -- Voltaire