|
![](/i/fill.gif) |
To Micha and ABX:
At risk of seeming offensive:
I couldn't care less about meshes - triangles are not my thing.
What I want is to trace a network of patches subject to a FFD, without
generating the approximation mesh.
To illustrate why it would be nice:
Design a blade of grass (using, say, 20 patches)
Apply a procedural FFD (calculated from the air flow)
Render it
Wow and drool at the smoothness and beauty of the movement
No difference yet between the two methods
Now, disperse the blades of grass over an HF, with a Poisson distr (just to
be a smartass), say, 100.000 of them
Let's see...
triangle meshes...what! *THAT* many *different* polygons??
One FFD per cloned blade of grass? Hmmmm...if i can find a compact way of
specifying the FFD this could be good. Oh. I can't apply a FFD to a network
of patches. Bummer.
I might take another route tho...model the single blade in F-rep, devise a
periodic version of the F-rep and superimpose a scalar field of parameters
to obtain local deformations...that sounds interesting too, well within SDL
domain.
However, a F-rep of that magnitude would be a daunting task to devise and I
would have appallingly little control on the distribution of single blades
of grass...
As of rendering time...anyone working on a parallel version of POV? Yes?
Good... Anyone could lend me for a few months a 512-node render farm? Make
those nodes at least R10000, please.....
Jokes aside, traced FFDs could help many mass scenes.
Alex
P.S.: I am truly amazed...I managed to start with area lights and arrive to
grass!
All hail the King of Offtopic!
Ha!
Post a reply to this message
|
![](/i/fill.gif) |