[MPlayer-dev-eng] comprehensive lavcopts test - in progress.

[Roman Gaufman]

Which is why I'm doing a PSNR vs filesize ratio :), when you work out the 
ratio, it becomes consistent and you can see if the gain in filesize is 
infact worth it or not. 

I'm still debating if pass2 encoding is a better test than vqscale.  Had a 
conversation on #mplayer, and it seems pass2 encoding is overall superior. 
Some opts wont work with vqscale, and the final encoding will be pass2, so 
whatever noise is added from ratecontrol should be taken into account for a 
fair test. As for ratecontrol deviation noise, why would it be random?

On Tuesday 15 June 2004 13:20, Jan Knutar wrote:
> > 1) PSNR ratio to filesize is also very important, not just PSNR
> > value, you should provide a difference in filesize as well (if any)
>
> For comparing PSNR ratio vs filesize, I think the only sane way is to
> use vqscale encoding, not 2 pass vbitrate encoding.
>
> It is my understanding that if you would compare file size AND psnr,
> WHILE using vbitrate 2 pass, all the adjusting of PSNR scores based on
> fielsize will just be random ratecontrol deviation noise.
>
> Could someone PLEASE explain this more clearer than I did? Or tell me if
> I'm wrong :-)
>
>
> _______________________________________________
> MPlayer-dev-eng mailing list
> MPlayer-dev-eng at mplayerhq.hu
> http://mplayerhq.hu/mailman/listinfo/mplayer-dev-eng