H264 Testing Part2
Well I’ve now briefed the MPlayer team on the issues and uploaded one of the clips. I also tried a full up-to-date debug build of MPlayer which I ran under GDB to produce this output. Sadly this build still crashes in the same fashion as my earlier tests.
A dig around on the FFMPEG mailing list produced an interesting posting with a patch that might fix the crash I’m currently getting -
--- mpegvideo.c~ 2007-10-25 05:42:50.000000000 -0700
+++ mpegvideo.c 2007-10-25 05:42:50.000000000 -0700
@@ -961,7 +961,7 @@
if(s->pict_type != I_TYPE && (s->last_picture_ptr==NULL ||
s->last_picture_ptr->data[0]==NULL) && !s->dropable){
av_log(avctx, AV_LOG_ERROR, "warning: first frame is no
keyframen");
assert(s->pict_type != B_TYPE); //these should have been
dropped if we don't have a reference
- goto alloc;
+ return -1;
}
assert(s->pict_type == I_TYPE || (s->last_picture_ptr &&
s->last_picture_ptr->data[0]));
After applying this I was able to play my test streams, but with lots of new and interesting errors. In particular the following appears most frequently.
[h264 @ 0x86a46b0]Interlaced pictures + spatial direct mode is not implemented
The video quality was less than perfect with lots of MPEG artefacts. A quick check of the FFMPEG mailing list shows that this appears to be a known problem with the current FFMPEG code and relates to the lack of “spatial direct” mode.
So perhaps two steps forward and only one step back at this stage. All of the streams do at least play now, albeit in a poor quality form.
Next step might be to try the CoreAVC H264 decoder with MPlayer to see if that works. At the moment I can only find instructions for version 1.3 of CoreAVC, whilst their website only had downloads of the most recent 1.6 release.