[Bf-committers] [Bf-blender-cvs] SVN commit: /data/svn/bf-blender [12455] branches/qdune/blender: Added support for subsurf orco coords, but still doesn't work properly

[Alfredo de Greef]

--- Joe Eagar <joeedh at gmail.com> wrote:
> Oh ok.  Well Geoff and me have been trying to figure
> out how to 
> calculate interpolation weights for blender's
> subsurfer for ngons; the interpolation code (e.g.
> for uvs and vertex 
> colors) is really weird and obfuscated.  If
> qdune's subsurfer supports ngons then maybe we can
> look at *its* 
> interpolation code. :)

I see, well, what you want to look into then are 'mean
value coordinates'. While there is actually code for
it in QD, it isn't actually used.
It was used only for the actual micropolygon value
interpolations and while it does produce superior
results, for mp's it was too slow. I haven't used it
yet for polygons/subsurf.
The subsurf code currently does this just by midpoint
subdivision, so in effect linear interpolations.

However, I happened to see that Brecht just a few days
ago commited code for exactly this purpose in arithb.c
So you can just use that instead :)
see
http://lists.blender.org/pipermail/bf-blender-cvs/2007-November/011714.html

Alfredo

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com