[Bf-committers] IRC meeting minutes, may 16, 2004

Jean-Luc Peurière bf-committers@blender.org
Wed, 19 May 2004 19:56:30 +0200

Le 19 mai 04, à 03:17, Matthew H Plough a écrit :

> You might want to look at the partial entity
> structure, described by S. Lee and K. Lee.  It is
> designed for nonmanifold meshes, and if I recall
> correctly, has roughly half the memory requirement
> of the half edge structure

I looked at the abstract, It seems it's half than the radial-edge
one which is derived of winged I think (Only from memory).

As the article is only accessible beyond ACM portal for which
I wont pay, I cannot check further.

In any case, all operations I spoke of previously works only on
2 manifold meshes, and as a non 2-manifold can always be
expressed as an array of 2 manifold meshes (even if decomp
are not unique), i think we can avoid the non manifolds structs
which are way more heavy to handle.

Now, If anyone has further view on that, I will be glad to hear,
almost all interesting extensions I found so far seems to need
direct edge support. Adding a partial support for each feature
seems also the thing to not do. If we can shrink the monster
editmesh in the same time, joy for all.