On Tue, 10 May 2005, Dik Takken wrote:

> One of my DV files reproducably triggers this error on last week's CVS 
> checkout:
        
        I don't think it's necessarily the DV data itself that is the
        problem - but rather the exact length (number of frames).

>     INFO: [mpeg2enc] Frame    49    49 P q=8.00 sum act=1886.01297
>     INFO: [yuvscaler] End of stream!
>     INFO: [yuvscaler] Normal exit: end of stream with frame number 50!
>     INFO: [yuvcorrect] End of stream!
>     INFO: [yuvcorrect] Normal exit: end of stream with frame number 50!
>     INFO: [mpeg2enc] Signaling last frame = 49
> mpeg2enc: seqencoder.cc:433: void SeqEncoder::EncodeStream(): Assertion 
> `pass1coded.size() == 0' failed.
> 
> Bug?

        Supposedly Andrew fixed that but I've seen it happen once in a while.
        But only on a Linux system.  On OS/X I don't see the problem.  It's
        a threading race condition with an OS dependency ;(

        What you can do to work around it is use "-M 0" which disables the
        threading logic completely in mpeg2enc.  The default of 1 just
        enables a 'readahead thread' but almost the identical effect can be
        achieved (if it is deemed necessary) with a buffering program (such
        as 'bfr' or even 'dd') in front of mpeg2enc.  On a local system
        (not pulling the data over a network) the default buffering in
        mpeg2enc is sufficient so extra readahead or buffering gains only a
        small improvement in performance.

        Values of of 2 or higher for -M achieve very little speed up - there
        are apparently enough serialization  points in mpeg2enc that threading
        doesn't speed things up much at all.

        There haven't been any changes to the encoder in a while but recently
        (yesterday and this morning ;)) the automake files and a few minor
        compile warning error fixes were checked in.  You might want to do
        a cvs update;./autogen.sh;./configure;make install when it is 
        convenient.

        Cheers,
        Steven Schultz



-------------------------------------------------------
This SF.Net email is sponsored by Oracle Space Sweepstakes
Want to be the first software developer in space?
Enter now for the Oracle Space Sweepstakes!
http://ads.osdn.com/?ad_id=7393&alloc_id=16281&op=click
_______________________________________________
Mjpeg-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mjpeg-users

Reply via email to