[Bf-committers] Idea for textureing method.

Robert Wenzlaff rwenzlaff at soylent-green.com
Wed Jan 5 15:19:23 CET 2005


I sat down the other night to make a tree (I know, "....only God...").

I made a very convincing bark texture from the procedurals.  I should say, it 
was very convincing on the trunk.  Where the branches went horizontal, it of 
course looked wrong.

Well this is typically where you would UV map.  But a tree with many branches 
all at differing heights doesn't unwrap well.  (I'm working on a character 
who is a squirel.  I'm going to need lots of trees.)   And it seems to be a 
lot of work for something as simple as "if the branch is horizontal 
(smoothly) sway the texture space coords to horizontal".

If it were a simple shape, like a bent tube, I'd use a lattice derform.  But 
thrying to lattice deform a cylinder into a tree would be a nightmare.

So I began thinking of ways to fill the gap between straight space XYZ texture 
and full UV mapping.

If we currently want an arbitrary mapping we insert an empty and use OB 
mapping.  What if the OB field were allowed to be a list of empties, or a 
special object type, and the resulting texture space was a distance weighted 
average of the empties' spaces?   You could set up a texture field where the 
XYZ mapped texture flowed following the empties.  Just as a vertex is 
effected by all the bones (in it's group), but the closest ones dominate.

If we don't want to clutter the texture screens with weighting and grouping 
controls, then a texture field could be a separate object type. (Think 
texture armatures).

It might also be a useful starting point for UV mapping.  You could "rough in" 
a texture with a field, then Bake it to a UV map.  You could even make a 
special pattern that could hint at how you want the mesh unwrapped.

My tree would simply have a few empties that followed the paths of the 
branches.  I could make twisting bark textures by rotating  the empties 
around the Z.
    
Thoughts?  Am I missing something?  I just had a thought that I could start 
with a hi-res lattice, deform it into a rough tree shape, and use that in the 
OB field of the texture (but not make it a deform parent).   A little 
trickier to edit than a field of empties, but it might serve the same 
purpose.  I'll test after I get back from work.
-- 
******************************************************
Why can't life be as easy as second-order homogeneous
 differential equations with constant coefficients?
******************************************************
Robert Wenzlaff        rwenzlaff at soylent-green.con



More information about the Bf-committers mailing list