[MPlayer-users] Re: [POOL] Moderate some users (was: Re: [-] Re: divx 6)

Rich Felker dalias at aerifal.cx
Wed May 10 20:05:10 CEST 2006


On Tue, May 09, 2006 at 03:32:03PM +0200, Matthias Wieser wrote:
> > > Regarding image quality it's at least an indication.
> >
> > I explained why it is not except in very limited cases. If your mind
> > is too small to grasp this then shut your mouth and go read some books
> > or run som experiments.
> 
> For most video codec parameters PSNR gives good indication if a changed 
> parameter results in higher or lower quality (bitrate, trellis, motion 
> search,..). For some parameters this is not true as they have more 
> influence on subjective image quality (quantization matrix, *_mask,...). I 

No, this is not what I was referring to at all. What I was referring
to is the fact that different types of artifacts have radically
different effects on PSNR relative to the damage they cause. PSNR
treats minor loss of sharpness the same as "25%" jpeg. You can call
this "psychovisual" if you want and claim that there're "objectively"
the same quality but this is idiotic. All it proves is that PSNR is a
stupid measure of quality. There are much better "objective" metrics
that show the difference clearly without bringing in any psychovisual
factors.

> But look what I have written one week ago:
> "But why do topics about the quality of a video codec so often result in 
> flame wars? Maybe it is because quality is a subjective, fuzzy measurement."
> Seems to be very true, even in the "Moderate some users" thread you start 
> your postings with insults.

Irrelevant. The problem is that you have NO technical knowledge about
codecs, nor the intelligence to realize that you have no knowledge and
that this makes it impossible for you to discuss these issues.

> I'm not interested anymore in discussing with you about wether you think 
> that gmc can or cannot help. I'm sure it CAN help sometimes, I have done 
> tests with different bitrates and even PSNR has increased.
> Since I have given actual numbers (contrary to you) and as I'm sure that 
> most people are convinced that there are videos where xvid's gmc can 
> increase quality, I don't see the point in spending more time with your 
> personal flame war.

On the contrary I think people know who knows their shit and who
doesn't.

> I have uploaded the videos so you can do your own subjective video quality 
> tests and try to prove that gmc made the quality worse (despite of the psnr 
> increase). You can find the videos here (available 10 days from now):
> http://service.gmx.net/mc/jFLZHaDQO6qFlpCxP4XDPJw5WOTrwX
> password: felker

I said ftp or http, not some nonsense rapidshare-like thing that
requires a gui browser, cookies, etc. I don't have time to waste on
anything I can't wget.

Also I am not interested in videos I have to encode to compare. I told
you that I will not do encoding; it's a waste of time. I don't even
have xvid and won't go thru the pain of trying to compile it. If you
want to show me videos you already encoded, be my guest.

Rich




More information about the MPlayer-users mailing list