Skip Talbot <Ski### [at] aolcom> wrote:
> It reminds me of my own attempts at clouds (see
> attached  picture).
Nice. They all appear to be very similar in shape, however. I take it the
only parameter being varied from cloud to cloud in your image was the bozo
translation, or similar?

> Your attempts are significantly better though thanks to the speed
> increases.
Gilles mentions on his site that this method is fast, presumably because the
density function is precalculated, but I was surprised at just how fast it
is. These clouds aren't very high-res dxf files, of course - 320x240x120
voxels - but the effect is all the more impressive because you just can't
tell.

>  The bozo pattern doesn't really hold up on the taller forms of
> cumulus and cumulonimbus.
True. I find that using a granite pattern instead of turbulence works well -
granite can produce the 'lumpy' appearance of clouds under the right
circumstances.

> You need to be able to model a cauliflower
> type shape and I'm still scratching my head on that one.  I thought
> about starting with a base shape like the gumdrop or hemisphere and
> tracing spheres of decreasing of radius on top of it, but haven't gotten
> around to actually playing with it.
Building a recursive merge of spheres and filling it with media sounds slow
to me... however, using such an object to define a pigment and hence a dxf
file would be much faster. Worth a try, at least.

I shall play more...

Bill

Post a reply to this message