On 04/26/2018 01:13 PM, Bald Eagle wrote:
> So, I was just looking around at some pattern functions, and having played with
> some Perlin noise in the recent past, thought that I'd read up on it a bit more
> - specifically it's exact implementation in code.
> 
> Does POV-Ray currently use original or improved noise?
> 
> http://flafla2.github.io/2014/08/09/perlinnoise.html
> 
> 
> Also, I came across this which touts a doubling of speed:
> Perhaps someone can test this out and see if POV-Ray can benefit from this.
> 
> http://riven8192.blogspot.com/2010/08/calculate-perlinnoise-twice-as-fast.html
> 
> 

Quite a long while ago I dug a 'little' into this and by the timing of 
Perlin's improved version I 'guessed' POV-Ray was probably using the 
original type, but I never verified it.

The difference IIRC was moving from a grid based approach to some 
simplex grid based one so the grid pattern was less visible in 2d 
applied noise. My initial pondering was along the lines of whether in 
3D applications of the noise the simplex grid would still be visible 
with the right alignments of surface to the new grid.

In the newsgroups I vaguely recall some folks (Christoph Horman?) 
commenting in some detail on perlin noise in comparison to POV-Ray's 
other methods. Perhaps search the newsgroups for more.

Bill P.

Post a reply to this message