Hi Rob,
I posted the ObjectToIso.inc file to the binary scene-files group a few days
ago, along with some example scenes. Perhaps I put it in the wrong place? The
technique in the SDL file is the same as the C++ one except much slower, so I
normally use the C++ patch of it.

By the way, I ran the Lucy mesh2 you provided with this technique - a nice test
- thank you. It ran in around 3 hours or so, no AA, single light source, but
with some gradient issues and I have never gone back to it.

It seems large meshes run as fast as even a relatively small collection of CSG
objects used as a single isosurface collection of equations.

I believe the reason is, throwing all the CSG objects into equations in the
isosurface defeats the very fast search mechanisms for objects that would
normally get used. It seems at least part of the fast searching on meshes is
still available to POVRay. Oddly this effect has me wondering about converting
large CSG count objects into single mesh2 objects on the fly...

It is in large part these new processors and POVRay's SMP implementation that
make the idea of using isosurfaces in this way something now doable. My longest
render thus far was the one of Bill Pragnell's archimedean solids. On an i7 920,
it came in at just under 4 hours using area lights, radiosity and AA. Though
this technique is very expensive, that kind of elapsed render time is workable.
This render would have been 30 hours plus on a single processor thread.

I believe too we have good room for improvements here, perhaps we'll figure
out how to go at this idea in better ways. I am myself still very much getting a
feel for it.

Regards, Bill

"Robert McGregor" <rob### [at] mcgregorfineartcom> wrote:
> "William Pokorny" <pokorny_epix_net> wrote:
> > Thank you Paolo, though I have to admit I was not aiming at any particular
> > effect, but rather I was testing the object to isosurface patch. I had time to
> > run another test last night - no idea what it is, but, dang, isosurfaces are
> > cool!
> >
> > The csg objects are on top, and the isosurfaces with some turbulence and a
> > little noise on the bottom.
> >
> > Bill
>
> This looks really cool Bill, but I must've missed something - I've never heard
> of ObjectToIso.inc - where's that available? I'm assuming it's a set of SDL
> macros that you've ported to C++ in a 3.7 beta patch, is that right?
>
> Cheers,
> Rob

Post a reply to this message