[Ffmpeg-cvslog] [ffmpeg]: r5416 - in /trunk:ffmpeg.c libavcodec/snow.c tests/regression.sh

[Luca Abeni]

Hi Corey,

On Sun, 2006-05-28 at 17:38 -0700, Corey Hickey wrote:
> lorenm at natsuki.mplayerhq.hu wrote:
> > Author: lorenm
> > Date: Sun May 28 14:38:10 2006
> > New Revision: 5416
> > 
> > Log:
> > Snow 1pass ratecontrol
> 
> This seems to work pretty nicely.
[...]
Can you provide an ffmpeg command line to test it?

I tried some encodings using an mpeg4 file encoded at 350kbps (test.avi)
as an input:
1) ffmpeg -t 60 -i test.avi -strict -2 -vcodec snow -b 50 out.avi
The bitrate of the encoded out.avi file resulted to be 684.8kbits/s
2) ffmpeg -t 60 -i test.avi -strict -2 -vcodec snow -b 600 out.avi
The bitrate of the encoded out.avi file resulted to be 657.3kbits/s
This looks good, but
3) ffmpeg -t 60 -i test.avi -strict -2 -vcodec snow -b 350 out.avi
The bitrate of the encoded out.avi file resulted to be 683.0kbits/s
Not good at all... Note that the original video is 350kbps.

Also note that in all the 3 cases I obtained about the same output
bitrate.

During the first encoding (the one with target 50bps), ffmpeg shows
q=31.0 all the time, so I tried
ffmpeg -t 60 -i test.avi -strict -2 -vcodec snow -qscale 31 out.avi
amd I got a 122.0kbits/s file. So, it seems that the ratecontrol is
trying to set the correct q value (ffmpeg shows that q varies - and the
q values seem correct), but the snow codec does not seem to be able to
properly respond to such variations...

What am I doing wrong? Or did I disunderstand the "1pass ratecontrol"
commit?

			Thanks,
				Luca
-- 
Proud to be "coglione"