[MPlayer-users] How to instruct mplayer to enable postprocessing with mpegpes? (NOVIRUS)

[D Richard Felker III]

On Fri, Dec 12, 2003 at 03:01:44PM +0300, Vladimir Mosgalin wrote:
> [Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
> On Thu, 11 Dec 2003, D Richard Felker III wrote:
> 
> DRFI>> In what way spp is better? As I see, it just smoothes video down to
> DRFI>> hell, removing artifacts in process.
> DRFI>
> DRFI>It uses the exported quantizer table (if available), and only will
> DRFI>"smooth video down to hell" if the table is unavailable, e.g. if
> DRFI>you do something stupid like applying another filter (such as pp)
> DRFI>first.
> 
> Of course I'm not doing that ;) Just something like -vf spp=3 or
> -vf spp=3,hqdn3d,noise=10uah,eq2 - the filter chain I use with spp
> instead of pp.
> 
> But results are blurred too much.

:/

> DRFI>> Sure, it may be useful if you have
> DRFI>> extremely low-quality video, but  something like pp=fq:15/hb:64:20/vb/dr
> DRFI>> with denoising afterwards will give similar results.
> DRFI>
> DRFI>The fq option should NEVER be used unless you have video from a
> DRFI>strange source that can't export quantizer tables. It will severely
> 
> Well, what should I do if I'm not satisfied with the pp's level of
> deblocking, and want to increase it? Strangely though, in windows

Use spp instead... :)

> DRFI>lower quality and increase blurring. The spp filter is much better,
> DRFI>provided you set the parameter high enough (e.g. 6).
> 
> Well, what cpu do you need to use level 6? ;) Mine 2ghz athlon can't do

A 386 will do, considering that I'd only use it for encoding and not
for realtime playback. Anyway the levels are very different. Low
levels only perform the dct/threshold/idct process at spread-out
intervals, whereas high ones perform it at every second pixel. I don't
know if using low levels could lead to blurring, but it will certainly
leave behind quantization artefacts that get translated around by
motion vectors.

Rich