Playback of joined Xvid-files on standalone

[Mike MU]

Hi!

I'm not sure whether this belongs in the XviD- or in the Standalone-forum...

A couple of days ago I started using XviD (v 1.1.0) instead of DivX, and I stumbled upon a problem I did not find any reference to in this forum or on the web.

When I paste two XviD-encoded files together (using virtualdub+append segment+direct stream copy), the resulting file plays fine on my computer, but my standalone (Targa DP-5100x, MediaTek MT1389FE, http://forum.doom9.org/showthread.php?p=581215#post581215) won't recognize the first frame of the second part as keyframe. It looks like only some macroblocks are recognized as being "key".
The same sometimes happens when jumping to a certain time, I can live with that though. It always disappears with the next keyframe.

This occurs with every source, profile, quant matrix and other setting I tried, it seems to be a rather general problem.

I never had this problem with DivX, and files that were encoded in one block with Xvid play back perfectly, too! (even with QPEL and eqm_v3ulr_rev3)

Why not use the "zones" in the XviD-Dialog?

- I can't define bitrate/targetsize separately for the zones (...right?)
- My main aim is to reencode only part of an encode to get rid of errors/artifacts apparent only there.

To be sure, I'll repeat the important part in short:
- keyframe not recognized at frame where files were joined
- only occurs with XviD, not DivX
- only occurs on standalone

Surely I'm not the first/only one with this problem...?

Edit:
Version:
XviD-1.1.0-30122005 _Final Release_

Based on CVS from 30.12.2005 20:00h MET

[eXistenZ_69]

Hi there,

maybe it's related to this:

http://forum.doom9.org/showthread.php?t=104156

You might want to read the MPEG4Modifier thread as well (http://forum.doom9.org/showthread.php?t=78050&page=13) from my post onwards. The latest version addresses this problem when unpacking...

Regards,

eXistenZ_69

[Mike MU]

Right on spot!
Both your links were very helpful, thanks a lot!

So it seems before pasting I always need to check the first part does not end with a B-VOP. Sounds logical. I can't help wondering though why playback on a PC is no problem...

[eXistenZ_69]

Glad I could help! :)

Regards,

eXistenZ_69

[Mike MU]

Just in case anyone with the same player read the above: I need to correct myself. Using another quant. matrix than h.263 sooner or later leads to faulty macroblocks. In the small sample I checked I didn't notice at first.