Am 13.02.2011 17:36, schrieb handos:

> While anti-aliasing can be done using the +A option but in my case it takes
> enormous amount of time because the scene is really complex. The other option is
> to create a super resolution image and then downsample it, but are there any
> other better ways to do it?

For anti-aliasing, there is /absolutely no/ substitute for oversampling, 
i.e. shooting multiple rays per final-image pixel. Whether you achieve 
that by (a) high-resolution rendering and subsequent downsampling, (b) 
any of POV-Ray's anti-aliasing modes, or (c) focal blur is up to you; at 
any rate, rendering at high resolution and then downsampling is always 
the most time-consuming variant of the three, for any given level of 
quality, as it will shoot the same number of rays for each final-image 
pixel, while the other variants will shoot fewer rays where 
anti-aliasing is not needed.

> I'm totally clueless about vignetting and bayer demosaicing. Is it possible with
> povray?

For vignetting, you could place a slab in front of the camera, with a 
texture that fades from totally transparent black at the center to 
almost-but-not-quite transparent black towards the edges.

As for demosaicing, I had to look up what that even is. Actually, the 
"root artefact" would be "Bayer mosaicing", with demosaicing being an 
attempt to compensate for that effect. For this, you'll probably need a 
postprocessing tool (though you might be able to use a second POV-Ray 
pass for this, using an orthographic scene). The following procedure 
might come close enough to the real thing:

Using a double-resolution output image, ...
- offset the R channel horizontally by 1 pixel
- offset the B channel vertically by 1 pixel
- average the G channel with a copy offset both horizontally and 
vertically by 1 pixel
- Downsample the image to 50%.

Post a reply to this message