POV-Ray : Newsgroups : povray.binaries.images : Compression experiment 3 (22k) : Re: Compression experiment 3 (22k) Server Time
17 Aug 2024 22:17:17 EDT (-0400)
  Re: Compression experiment 3 (22k)  
From: Kari Kivisalo
Date: 17 Aug 2001 09:30:25
Message: <3B7D1D1D.D16D6E2E@pp.htv.fi>
Fabien Mosen wrote:
>
> Ideally, we could also have render-time filters, when the filtering
> can be done on a pixel-by-pixel basis.

This would be good also because anti aliasing would work properly.
That room3 image was actually rendered with +am1 +r2 +a0.0 and
it didn't do much good. It's now clear that aa is best done on
final display (which is close to perceptually uniform space)
image because it's there for human vision. Doing aa on linear
image is just a waste of resources. So, I should have rendered
at 2x or 3x size without aa and resampled down as the final step.
 
> Would 24-bit depth create banding after compression ?

What I see and what I would expect to see in a reasonably realistic
image is quantisation noise. Banding maybe on perfecly flat surfaces.
This is how a 24 bit image looks after the process. The bright parts
are noise free.

http://www.pp.htv.fi/kkivisal/room3_24.png.


_____________
Kari Kivisalo


Post a reply to this message

Copyright 2003-2023 Persistence of Vision Raytracer Pty. Ltd.