gregjohn nous illumina en ce 2008-08-25 22:28 -->
> Alain <ele### [at] netscapenet> wrote:
> 
>> When randomly exploring a fractal, it's very hard to detect
>> if a given "location" is interesting before you actualy
>> render that location. Your test may flag a marvalous
>> location as uninteresting and skip it, and then render 100 of
>> extremely boring ones accidentaly flaged as "interesting"...
>>
> 
> No, with eval_pigment, one may compare the colors of two close-by points in
> fractal space.  If they are identical, then there's a good chance you're out in
> a field of blanket color.  I had a pretty good run of looking at fields where
> the colors were dissimilar. Not all worthy of uploading to zazzle, but
> certainly interesting.    I think the most interesting test (but biggest
> longshot as far as remote odds) would be to look for a black spot and a
> non-black spot in the same field.
> 
> 
You'll need to sample 100's of pairs to be sure. If you sample only 2 locations, 
you may well miss a very interesting image. Two close-by points can be 
identical, and be just next to some very busy area, or just apen to lie 
parallel, or almost, to the local gradient.
You need a bare minimum of 3 points in a triangle.

-- 
Alain
-------------------------------------------------
You know you've been raytracing too long when you look at real clouds and 
criticise their media and radiosity settings.
Tom Melly

Post a reply to this message