[MPlayer-users] lanczos or bicubic spline or..?

[D Richard Felker III]

On Sun, Feb 01, 2004 at 11:11:05PM -0800, * wrote:
> Thanks (again) Rich!  I think I'm a little less confused now.  I don't
> know if this is complete nonsense or not, but I ran a test upscaling 10
> times in a row to compare:
> 
> mplayer movie.vob -nosound -vo png -vf crop=716:452:0:12,
> scale=720:384,scale=736:400,scale=768:416,scale=800:432,scale=832:448,
> scale=864:464,scale=896:480,scale=912:496,scale=944:512,scale=976:528
> -sws #

IMO this is not a legitimate test. When encoding a movie you don't
scale 8 times, and you especially should never upscale. Try it with
just _one_ scale filter, to the size you'll actually be encoding, and
compare the results.

Also, compare the results _after_ encoding to mpeg4, since that's what
will ultimately matter.

> http://box.dnsalias.net/rose/orig.jpg
> http://box.dnsalias.net/rose/0.jpg
> http://box.dnsalias.net/rose/1.jpg
> http://box.dnsalias.net/rose/2.jpg
> http://box.dnsalias.net/rose/3.jpg
> http://box.dnsalias.net/rose/4.jpg
> http://box.dnsalias.net/rose/5.jpg
> http://box.dnsalias.net/rose/6.jpg
> http://box.dnsalias.net/rose/7.jpg
> http://box.dnsalias.net/rose/8.jpg
> http://box.dnsalias.net/rose/9.jpg
> http://box.dnsalias.net/rose/10.jpg
> 
> I see that the images with more blur have a higher PSNR. 

Huh? Why do we have jpegs here? And how did you measure PSNR?

The PSNR I'm talking about is not from the scale filters, but from the
mpeg4-encoder after the scaling takes place. What I said is that if
you give the mpeg4 encoder a sharper image, your resulting movie file
will have worse (lower) PSNR relative to the input it gets, than if
you give the mpeg4 encoder a more blurred image.

> Which is not
> what I want (well, at least not 10 times in a row).  To me, the only
> images that look acceptably sharp are 2, 6, 9 and 10.  And 10 has some
> horizontal banding.  9 (lanczos) seems just as sharp as the original,
> except maybe the contrast is higher, or it is grainy.

0, 3, 4, 5, and 8 definitely suck. They're basically useless, except 8
(sinc) has the nice property that if you upscale then downscale, you
should get back the original image almost perfectly.

1 and 7 will blur somewhat, but that's sometimes good.

IMO you should pick between 2 and 6. 6 probably makes more sense for
4:2:0-sampled sources.

> I should also mention that I've been using the hqdn3d filter based on
> your recommendation- it works amazingly well, thanks!  But in my quest
> for fine details I've turned off the luma and chroma blurring
> (hqdn3d=0:0:6).  I hope there is nothing wrong with this.  Thanks.

Well it won't help as much that way. I'd set them to something low but
nonzero....

Rich