Also sprach Ari Huttunen zu "02.12.2003 12:25" Anno Domini:
>
> Is this the same 'quantization' that is limited by ffmpeg's "-qmax 10" that
> I have there? If so, any idea what amount of quantization generally results
> in macroblocks being visible? Or does this question make no sense?
>
Looking shortly over ffmpeg.c, I saw that the default value fo option
"-qmax" is 31! That's quiet more than you set, maybe this is a
explaination. Why do you set it explicitly, and why to such a low value
(it is the allowed maximum, meaning a upper bound). Try using default,
by not specifying the option!
Overall, I would suggest, to leave as most options as possible at their
default values, which most often are the best choise for standard tasks.
This is also true for other applications like mencoder!
> I'll see what happens if I use mencoder. I first used that one, but I found
> it harder to specify that I want to cut out "N pixels from the left", which
> is useful for some recordings.
>
Make sure, your output resolution is divisible by 8 in height _and_
width. Otherwise, you will make the encoder crazy, as its algorithem
depends on 8x8 macroblocks! There is a quiet good tool in
mplayer/mencoder: You can use "mplayer -vop cropdetect ..." and mplayer
will try to find a best crop area. The result is printed on the console,
ready to use as option for encoding. You can even use "mplayeer -vop
rectangle=<crop-parameters> ..." to verify (and even correct) the
result. Read mplayer's/mencoder's HTML documentation for mor infos!
--
Patrick Cernko | mailto:errror@errror.de | http://www.errror.de
Quote of the Week: "/vmlinuz does not exist.
Installing from scratch, eh?"
(Debian Kernel Installation)
Attachment:
pgp00005.pgp
Description: PGP signature