[MPlayer-users] lavc vs. xvid (and improving lavc quality)

[Matthias Wieser]

rcooley wrote:
> On Sun, 6 Jun 2004 13:33:10 +0200
>
> matthias-wieser at t-online.de (Matthias Wieser) wrote:
> > ??? It doesn't matter if you use 1 or 2-pass encoding if the problem
> > lies at the strategy how mencoder distributes the availabele bits
> > within one frame.
>
> That is not the problem here.

I think it is.


> > You may try to explain how to do 2-pass encoding with TV-sources.
>
> First you save it to a lossless format, and then you do two-pass
> encoding on the resulting file.

haha.

> I never said you should do 2-pass on a TV source, just that
> single-pass can't give very good results without very high bitrates.

Single pass gives for 95% of all cases good enough results. The problem I 
describe is not related to single/dual-pass, bitrate or any other option.


> > denoise3d doesn't lower quality of tv broadcasts significantly. It
> > improves compressibility.
>
> Yes, denoise does lower picture quality plenty.

Then please file a bug report is denoise3d or hqdn3d doesn't work for you. 
I even have done some comparisons. Maybe you want to read
www.wieser-web.de/ or
http://home.knuut.de/MWieser_/vf_raw_film_noise2_3/


> Who said you need a new CPU?  They aren't all CPU-intensive options,
> quite a lot of them don't affect CPU time at all, they just change how
> lavc allocates the bits it has to work with.

The problem is not temporal distribution of the bits but local 
distribution.