[Mplayer-cvslog] CVS: main mencoder.c,1.95,1.96 cfg-mencoder.h,1.27,1.28

[Michael Niedermayer]

Hi

sorry for the delayed awnser(s) but the gmx smpt server seems to delay my 
mails for a few hours or days? ;)
so i send this one again, over a more stable smpt server :) sorry for the 
dupplicate msg, but i thought ill explain why some of my awnsers seem so slow 
since yesterday ...

i think all other mails did end up on the lists allready ...
-------------------------------------
Hi

On Monday 18 March 2002 20:00, Gabucino wrote:
> > Gabu will do better comparison (i hope) - he has big experience in such
> > tests...
>
> Ah :) I wouldn't dare to say that ;)
> Spent this day testing divx4 vs mpeg4, and I couldn't come to a clear
> conclusion (yet). However, this indecisiveness means that I can't
> decide which one's quality is better. Some remarks :
>  - uploaded some sample png's to http://mplayerhq.hu/~gabucino/test/mpeg4
>    I'll refer to them below.
>    I have very limited net, sorry :( /see next remark/
>  - It seems to me that lavc mpeg4 encodes with too high bitrate in the
>    beginning of the file (and can't set smaller), so can't test reliably
yes, its a bug, allthough i dunno how to fix it, as without that it will 
create very bad looking stuff if there is high motion at the beginning ...

[...]
>  - I didn't use any other libavcodec options, just vbitrate - as I don't
>    know them..
>  - My opinions may change unexpectedly! ;)
>
> I agree in what we already know: lavc mpeg4 beats divx4 by its
> speed/quality value on high(er) bitrates. (mostly checked on lower
> bitrates, yet..)
iam surprised that lavc can beat divx4 at all, with the crap motion 
estimation it has at the moment

>
> I'll make some more tests tomorrow - I'm not sure about my current
> opinion :))
perhaps u want to wait until the motion estimation is fixed ... could take 
some time though 

btw. thanks for the comparision :)

Michael