video compression error

[kilg0r3]

Hi there,

I've got a problem with koepi's 02052002-1 release. When i try to compress a clip in quantizer or quality mode. After about 2550 frames virtualdub (1.4.9 or 1.4.10) produces the following error.

video compression error: an unknown error occurred (may be corrupt video data)

Dvd2Avi stopped at 3024.

used settings were: quality 20%, quantization type mpeg, rest was default.

[sffubs]

Have you tried any of the other binaries (e.g. Nic's or uManiac's)? You never know, this might be specific to the bins you have.

-s

[Foo]

Originally posted by kilg0r3
Hi there,

I've got a problem with koepi's 02052002-1 release. When i try to compress a clip in quantizer or quality mode. After about 2550 frames virtualdub (1.4.9 or 1.4.10) produces the following error.

video compression error: an unknown error occurred (may be corrupt video data)

Dvd2Avi stopped at 3024.

used settings were: quality 20%, quantization type mpeg, rest was default.

Do you have MV hints checked? This is an old issue. Not really sure why quantizer mode looks for the MV file, but if you have it checked while trying to encode in constant quantizer mode, there will be a memory leak until Vdub errors out. If you go in and uncheck the option to use MV hints (can't do it directly from quantizer mode so switch to a mode where you can) it should fix the problem.

foo

[Koepi]

Uh, nice that you tell me, didn't found that this is the solution for it.
I can change that with my next binary.

Thanks,
Koepi

[manono]

And it's not specific to Koepi's versions. It happens with Nic's as well. But making sure that HME isn't checked is the work around.

[Foo]

Originally posted by Koepi
Uh, nice that you tell me, didn't found that this is the solution for it.
I can change that with my next binary.

Thanks,
Koepi

Sorry :D . Someone else brought this up in another thread a while back and asked that it be fixed. Being the _patient_ person I am, I figured you or Nic or -h had seen the request and would get around to it ;) . Thanks for the fix whenever it comes.

foo

[Nic]

(I must admit I know i'd seen the bug report...& im almost 100% -h had seen it too.....memory leaks are a programmer's nightmare :) . If Koepi doesnt get chance ill try it)

-Nic

[Koepi]

Ahoy,

ok, I have fixed this issue.

Sadly, Nic's DSF doesn't work with the actual core and the XviD-DSF itself doesn't compile either on my machine, so I can't release a binary yet :-(

So let's wait for Nic to release a binary (I sent him the modified sources), so I can fetch the updated DSF from him and release my package as well.

Best regards,
Koepi

[Koepi]

Ah, Nic downloaded the sources already, so it should be short time until the new binary release.

I finally managed to get the original DSF to compile (screwed up my DX SDK somehow, well, finally all paths are fixed again ;) ), but I think Nic's DSF is better since it has postprocessing, so I'll wait until then.

[Koepi]

OMG,

I did a little further investigation and found possibly an error that screwed up hinted MVs...

Strange that we didn't notice any errors there.

I'll check around a little and see if this is really a problem ( an & instead of a && is a bad thing in c / c++....)

L8er,
Koepi

[kilg0r3]

Oooops,

and i thought it was all my fault :D.

cheers to you all

[Koepi]

Heh, it wasn't all too bad, I just misread some lines in the code, everything is fine.

But with the new builds from me you shouldn't experience those crashes anymore. Maybe foxer will check this in into CVS, I still don't have CVS access :-/

Regards,
Koepi

[Psyche]

@Koepi:
I've open a topic on this in xvid forum, where I give an alternative (and IMHO better) fix to this error.
BTW, now that -h is gone for a couple of months we need somebody to make changes in vfw interface while he can't. And the only person who comes to my mind is !Koepi!, what do you say?
You, after -h, are the person who knows xvid vfw interface better, why don't you get rw access to cvs? Maybe exams?