[MEncoder-users] B-frames and GOP with MPEG-2 for DVD

[nicolas.hesler at sheridanc.on.ca]

> Nicolas Hesler wrote:
> 
> >BUT, I've got another problem (that it appears I've had all along 
> with 
> >mencoder encodes), which I've only just discovered and that is 
> that I 
> >have pulsating (grain/artifacts) frames at what I'm guessing is 
> every 
> >keyint.   It's not noticeable on every video I've encoded but some 
> >become unwatchable. I've tried encoding with b frames and closed 
> gop, 
> >but to no avail.   I've tried simplifying my command line to make 
> >testing for manageable but had no luck with
> >
> >( mencoder -oac lavc -ovc lavc -of mpeg -mpegopts 
> >format=mpeg2:muxrate=10080:tsaf:telecine
> >-vf scale=720:480,harddup -srate 48000 -af lavcresample=48000:0:2
> >-lavcopts 
> >vcodec=mpeg2video:vrc_buf_size=1835:vrc_maxrate=7000:vbitrate=6000:keyint=12:
> >trell:mbd=2:vpass=1:acodec=ac3:abitrate=224:aspect=16/9:vstrict=0 -
> ofps 
> >24000/1001 -o $output $input)
> >
> >Any ideas on how to resolve the pulsating???
> >
> >current command line:
> >
> >mencoder -oac lavc -ovc lavc -of mpeg -mpegopts 
> >format=mpeg2:muxrate=10080:tsaf:telecine
> >-vf scale=720:480,harddup -srate 48000 -af lavcresample=48000:0:2
> >-lavcopts 
> >vcodec=mpeg2video:vrc_buf_size=1835:vrc_maxrate=7000:vbitrate=6000:keyint=12:
> >trell:mbd=2:precmp=2:subcmp=2:cmp=2:dia=-10:predia=-
> 10:cbp:mv0:vqmin=1:lmin=1:dc=10:>vpass=1:acodec=ac3:abitrate=224:aspect=16/9:vstrict=0 -ofps 24000/1001 
> >-o $output $input
> >  
> >
> 
> vstrict should be set to 1 and dc=10 looks very suspicious to me 
> when 
> targeting dvd.
> Someone suggested some variation on vi_qfactor and vi_qoffset to 
> reduce 
> the pulsating effect,
> but unfortunately I don't remember the values he adviced.
> Another parameter I always use is sc_threshold=-1500 to increase 
> lavc's 
> sensitivity
> to scene changes


I've tried setting the vi_qfactor and vi_qoffset  suggested here:
http://lists.mplayerhq.hu/pipermail/ffmpeg-user/2006-April/002770.html

-lavcopts vi_qfactor=0.1:vi_qoffset=1.5:

but it didn't improve the pulsating or "pumping" as some people have described it elsewhere on the keyframes (keyint).  I've changed the keyint parameter a few times to confirm that the pulsating coincided in frequency and it does.

Can anyone suggest which parameters might have some influence over the pulsating on keyframes?  Are the keyframes being compressed differently so much so that they "pulsate"?  Any ideas on what I should experiment with to begin more tests?  

Note: My source content is cartoon, is 24 fps and soft-telecined, and the pulsating is most visible in darker areas but that is probably just a perceptual thing.