[Bf-committers] Booleans

Theodore Schundler tschundler at scu.edu
Thu Sep 9 10:38:54 CEST 2004



My plan is to get it fully functional in python first (except vertex
colors & UV coordinates because those are trivial) to test the
algorthims used. Once it is functioning to satisfaction (hopefully by
the end of this month), I plan to re-implement the algorithm in C (or
maybe C++) as a replacement to the current booleans. So that would
tentatively be rea-dy around mid october. Additionaly, I'm planning to
make the C version available to python scripts, and maybe as a non
distructive effect since people like the idea of using in animations and
visualizations and such.
The reason for starting in python is because testing the algorthms is a
lot easier (Alt+P vs. quit,recompile,run,test) and anything
experimenting with an idea like this is going to need a rewrite anyway.


Thanks bart for the link to the b.org forum post. Looking at Laurence's
improved booleans, they are an improvment. But it looks like they
fundamentally work the same way as the old ones, subdividing the mesh as
they go. That still leads to extra verticies and faces. My method works
in passes. First it finds all points of edges intersecting with faces
nad detemines what is "inside" or "outside". Then it calculates the new
edges along the intersection. Then as the final step is adding the faces
to the new mesh. The faces where there are intersections are effectively
n-gons after the operation (sorry for using the n word) and get filled
(previously using Blender's scanfill, but now I'm working on filling
with quads, the current version on elysiun uses a messy mix of both that
will be cleaned up in my next release next week).

Since my method produces cleaner results and is more extensible (i.e.
can use 2D shapes as cutting tools and potentially can support making
smooth faces, not just flat ones) I plan to deffinately continue
development.

>From my guesses from the results produced by the one posted on b.org's
forums, Laurence and I are using quite different approaches. So
unfortunately, I doubt his code can be used to improve my system much,
and similarly, my code won't be much useful for improving his. (Unlike
Jeorge (toloban on Elsyiun)'s method which works similar enough that I
can intergate his method for quad filling into mine.)


Ted

>>> ton at blender.org 09/08/04 3:57 AM >>>
Hi,

Work on that area is definitely welcome! What is your proposed next  
step? Make this a C library, stick to python, or recode the current  
system?

-Ton-



This message scanned for viruses and SPAM by GWGuardian at SCU (MGW1)


More information about the Bf-committers mailing list