Analyse B-frames for Xcard

[Jvdhorst]

Owning a X-card I have problems playing Xvid 1.0 clips containing b-frames. They stutter. Looking at the movements in the clip when this occurs I suppose it happens when bframes come by. No matter how many bframes or whatever the packed bitstream or closed GOV setting it can not be solved. Sigmadesigns, the creators of the card, will not comment on the issue. Somebody once said that sigma claimed that the xvid stream is non mpeg4 compliant but I can't find that statement anywhere nor will they repeat it. Xvid is the only codec I found so far showing this stutter.

I would like to find out myself why exactly the clips won't play smoothly. I want to do this by comparing the video streams from other codecs containing bframes that do play smoothly. I want to compare the way b-frames are written in the bitstreams and perhaps develop a little tool to convert. Would someone be able to tell me how to analyse an avi's mpeg4 stream directly? Or is there anyone with extensive experience willing to help?

If you're interested in the research I already did check this thread:
http://www.sigmadesigns.com/dcforum/DCForumID7/808.html

[Koepi]

A search on this forum would have brought you to some interesting threads, unfortunately they're not the search default 30 days away but longer.

There we had discussed which driver version can cope with which xvid version or something like that. You might want to refine your search with this knowledge.

Regards
Koepi

[Jvdhorst]

I know this, and in the link to the thread I provided you can find that I've investigated that. Playback is smooth until you go ahaed and FF or RW or pause. Then you lose lipsync. Sometimes there is no lipsync from the start. Also the older drivers have some other annoying issues. So downgrading the drivers is not an option. I have probably followed every discussion out there about the issue, but it's still not known what exactly the cause is.

[SeeMoreDigital]

Hi Koepi,

I'm with Jvdhorst here.

I've tried every single version of XviD (since Dec 03) and it does not matter if you use 1B-VOP, 2B-VOP or more. With packed bit-stream or without packed bit-stream. XviD B-VOP will not work with the Xcard!

I've tried using GraphEdit and 3ivX's muxer to output .MP4 files. I've tried changing XviD's 4CC, MPEG4Modifier to change the encodes UserData... but nothing works.

Until recently we Xcard users could only test our hardware cards using DivX's 1B-VOP implementation. However, that's all changed because we can now use Recode2's 1B-VOP encoder. Which works perfectly!

So what makes XviD's B-VOP implementation so different from DivX's and NeroDigital's?

Is there a bit of hidden XviD code that sends the Xcard into 'turd' mode every-time an XviD encode is played.

Can you offer us any information please?


Thanks

[EnF]

Owning a X-card I have problems playing Xvid 1.0 clips containing b-frames. They stutter. Looking at the movements in the clip when this occurs I suppose it happens when bframes come by.

Same here.

I've recently tried Xvid (1.0.2-29082004) - an encode with default settings (2 max. consecutive B-VOPs).
I've noticed that the stutter appears mostly after scene changes.
Usually that's the place where an I (key) frame is introduced, right?
Could it be a combination of I and B frames?
It doesn't stutter all the time, sometime there are couple of seconds smoothly played but I'm sure there are B-VOPs in those frames as well.

It happens even with DivX 5.1.1 (just 1 consecutive B-VOP) although I know there are people arguing that. (again, immediately after scene changes)
The stuttering is not as noticeable as with XVid but it is there.

[SeeMoreDigital]

Hi Enf,

The XviD B-VOP situation has not changed even with the proposed XviD-1.1.-127-13102004 (which is in beta at the moment).

I too have noticed problems with DivX5.1.x and DivX5.2.x. I think this has something to do with their new "adaptive" seaking/implementation... But what ever it is it ain't good :(

About the only Mpeg4 codec with B-VOP that works now is Ahead's Recode2. Unless you go back to pre-DivX5.1.0 builds!


Cheers

[Jvdhorst]

Pfoe long time since this discussion started. Anyways I think it's safe to conclude now that the problem is with the Sigma drivers, specificly it's own avi parser. A possible solution would be making our own parser, but I'm not experienced at all with directshow or low level specs of the avi file format. If somebody would be able to direct me to some good resources about the the subjects I could give it a shot. I'm experienced in programming in C++ and programming parallel processors, but only in the application of scientific image processing (which is very different from multimedia control applications).

[SeeMoreDigital]

Originally posted by Jvdhorst
.. Anyways I think it's safe to conclude now that the problem is with the Sigma drivers, specificly it's own avi parser... I wonder if, as a temporary measure, we could test the Xcard with somebody elses parser?

That said... I wonder how many AVI parsers are there... out there?


Cheers

[SeeMoreDigital]

I've just had another thought...

How can XviD's B-VOP problem be related to Sigma's AVI parser, when the streams don't work when muxed into the .MP4 container either?


Cheers