[Ffmpeg-cvslog] r7132 - trunk/libavcodec/mpegvideo.c

[Baptiste Coudurier]

Hi

michael wrote:
> Author: michael
> Date: Mon Nov 20 11:48:23 2006
> New Revision: 7132
> 
> Modified:
>    trunk/libavcodec/mpegvideo.c
> 
> Log:
> 100l qprd fix 2nd try
> 

Ok, Tried a few more times. I still experience high (weird?) q values.

/usr/local/bin/ffmpeg -y -i test_motion_est.avi -flags +qprd -mbd rd
-lmin 1 -mblmin 1 -qmin 1 -b 10000k -bf 2 test.vob

See attachments for q values.

all 60... I dunno if that's rc default behaviour to raise so much q
value for B frames. I can greatly improve encoding by limiting qmax to
reasonable value.

Is there an option Im missing to avoid that besides qmax/lmax ? Or is it
the sample encoded that causes that rc wierd behaviour ?
Mainconcept encoder seems to adopt the same I/P behaviour for q in B frames.

Atm I can see really visible artefacts on B frames, near the number at
the bottom of the picture.

Also I still confirm the infinite loop with:

/usr/local/bin/ffmpeg -y -i test_motion_est.avi -flags +qprd -mbd rd
-lmin 1 -mblmin 1 -qmin 2 -b 10000k -minrate 10000k -maxrate 10000k
-bufsize 2000k -qmax 6 -bf 2 test.vob

-- 
Baptiste COUDURIER                              GnuPG Key Id: 0x5C1ABAAA
SMARTJOG S.A.                                    http://www.smartjog.com
Key fingerprint                 8D77134D20CC9220201FC5DB0AC9325C5C1ABAAA
Phone: +33 1 49966312


-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: qprd_q_values.txt
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-cvslog/attachments/20061121/36e1d67a/attachment.txt>