I'm kind of wishing I'd tried this earlier. The voronoi pattern (which 
I've posted about before and included in the latest MegaPOVPlus release) 
uses a kd-tree data structure similar to that used for photons. My 
implementation is very generic, and allows other data to be included.

Anyways...take a very slow pattern, and sample it at many points. Dump 
those points and values in a kd-tree, and at render time, collect the 
points in a given radius around each point and average them together 
weighted by some falloff function. The result renders very fast, and 
smoothes out grainy appearance. You can determine an acceptable 
compromise between grainyness and detail by adjusting the radius for 
gathering samples. It can also give a more efficient way of storing a 
voxel map, especially when only a relatively small proportion of the map 
is "filled", since only the voxels actually used are stored.

This image took 30 seconds to render on a 933MHz PowerPC G4. The pattern 
map construction time is a bigger issue: about 6 minutes for half a 
million points with the settings and algorithm I used. However, since 
this is done in a preprocessing step, it is still fast enough to use 
with focal blur, media, etc, and increasing the image size has a 
relatively small influence on total render time. Without the pattern 
map, I estimate it would take at least 15-30 minutes, and would still 
appear very grainy.

I have made another pattern (another interface for the same one, really) 
called the "smudge" pattern which allows you to specify the individual 
points with colors, pigments, etc, the result being somewhat like the 
blob pattern. It is something like spray painting or airbrushing. Like 
the voronoi pattern, it obviously isn't terribly useful for direct hand 
coding, but some interesting effects are possible using various 
algorithms to generate the points.

-- 
Christopher James Huff <cja### [at] earthlinknet>
http://home.earthlink.net/~cjameshuff/
POV-Ray TAG: <chr### [at] tagpovrayorg>
http://tag.povray.org/

Post a reply to this message

Attachments:
Download 'smudge.png' (38 KB)

Preview of image 'smudge.png'