POV-Ray : Newsgroups : povray.binaries.images : Stochastic render rig focal blur FIXED : Re: Stochastic render rig focal blur FIXED Server Time
1 Aug 2024 16:24:11 EDT (-0400)
  Re: Stochastic render rig focal blur FIXED  
From: Edouard Poor
Date: 23 Aug 2008 05:05:44
Message: <48afd2e8@news.povray.org>
Chambers wrote:
> I'd like to know, what is the difference between this and POV-Ray's inbuilt
> focal blur?
> 
> Shouldn't the POV blur be faster, since it can use the adaptive level to skip
> some samples?
> 
> And if this yields a better result than the inbuilt, how much work would it be
> to implement this version instead of POV's current version?
> 
> ....Chambers

POV-Ray's focal blur is certainly faster than this technique - although 
this method can give you shaped bokeh effect.

Where this method is sometimes a win is when you are using several 
techniques together - for example I want to render some scenes that have 
the following features:

   - Anti-aliasing
   - Focal Blur
   - High density light dome (100-1000 area lights sampled from an HDR 
lightprobe)
   - Blurred reflection and refraction

Normally to do that you'd turn on AA and DoF, create all the lights 
(say, 200 area lights), and use multi-texturing to average together a 
large number (e.g. 256) of micro-facets on the blurred reflector and 
refractor objects. The render times are huge in that case.

Instead of rendering a single image with all that turned on, I instead 
render a large number of frames (anything from 50 to 1000 or even more) 
with no AA, no DoF, one random 2x2 area light, and one random 
micro-facet texture. Each frame renders quite quickly in that case.

Each frame has a different shear factor that, when averaged together, 
gives a depth of field focal blur, a slight jitter of the camera 
position to give (when averaged) anti-aliasing, as different light 
source, so that the final averaged image has all the lights present, and 
  a different micro-facet on every blurry surface to give a proper blur, 
again when all the images are averaged together.

So it isn't useful to just replace a single feature, since POV-Ray 
usually has adaptive techniques that would be faster, but it is useful 
to reduce the render times of scenes with a particular combination of 
features (like in the list above).

I really must come up with an example scene to try and show this off...

Cheers,
Edouard.


Post a reply to this message

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