On Sat, 6 Dec 2003, Andrew Stevens wrote:
> > If you'd still like a piece of the capture, I'll gladly push some
> > stuff your way. But, in further experimenting tonight I think I've
> > found the culprit. If I drop the -Q 4.0 parameter, then rc92 stops
Richard uploaded a 70MB snippet for me and I just now took a look
at it with similar (but not 100% identical) encoding parameters (but
I did leave in the high setting of -Q).
No artifacting that I can see - and I was looking for the same
scene that the initial .jpg was created from.
> There were some changes in bitrate control in rc92 and it could be that with
> the new bitrate control regime -Q 4.0 is enough (for this particular
> sequence) to kick off some blocks where the MMX DCT routines run out of
> nuermic precision.
Possible that it's the SSE DCT that's having trouble instead of the
MMX one? The encoding I just did was on a system that doesn't
have SSE support (the old MMX vs SSE problem we've discussed in the
past) and using the current cvs version of the encoder I can't
reproduce the problem yet. Time to go try a different system.
> PS. -Q 4.0 is really really high!!!
That's what I thought too.
Cheers,
Steven Schultz
-------------------------------------------------------
This SF.net email is sponsored by: IBM Linux Tutorials.
Become an expert in LINUX or just sharpen your skills. Sign up for IBM's
Free Linux Tutorials. Learn everything from the bash shell to sys admin.
Click now! http://ads.osdn.com/?ad_id=1278&alloc_id=3371&op=click
_______________________________________________
Mjpeg-users mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/mjpeg-users
