[Bf-committers] Mesh curvature estimation

[Vincent Nivoliers]

There was one in 2010, here is the page of the project :
http://wiki.blender.org/index.php/User:Rohith291991/Gsoc2010/Proposal

I spent a bit of time discussing about it with Rohith on IRC, and provided some code I had written for a partial implementation of mixed integer quadrangulations into an other software. Last time we discussed about it about one year ago, he was struggling to extract a mesh from the parametrization. Yet, I don't know if the direction field was still manually generated, but I think he used the code I gave him to build the parametrization from a direction field. I'll try to go over his code to see if I can guess how far he went.

Anyway, I don't know what's left of this project, and what can be recovered. Still, if you want to resurrect this project, and think I could be of any help, please let me know, and I'll send you my code as well, and we can arrange a meeting on IRC.
-- Vincent.

----- Mail original -----
> I dont think the MIQ project got very far, since I've offered to
> implement this once my work on compositor / mask editor is done, and
> my offer was generally greeted with much happiness that "someone" was
> going to do this...
> 
> noone even mentioned to me that anyone did a gsoc on it... so im
> guessing it didn't go TOO far... but maybe I'm wrong... idk.
> 
> On Mon, Nov 28, 2011 at 6:01 AM, Vincent Nivoliers
> <vincent.nivoliers at loria.fr> wrote:
> > [2011-11-27 19:38:53 -0500] Nicholas Bishop:
> >> Hi all,
> >>
> >> A few times I've run into algorithms that require principal
> >> curvatures
> >> [1] as inputs. My cursory examination of the Blender source
> >> (basically
> >> just grepping for "curvature") didn't turn up any code for this, so
> >> I
> >> ported the (GPLv2) curvature code from the trimesh2 lib [2] to
> >> Blender. That code is based on the paper "Estimating Curvatures and
> >> Their Derivatives on Triangle Meshes" [3]. Dunno if this is useful
> >> to
> >> anyone else, but I've pushed a Blender branch [4] containing the
> >> curvature calculation (in blenlib/intern/curvature.c), as well as a
> >> test modifier that displays the curvature vectors.
> >>
> >> Screenshot with Suzanne:
> >> http://www.pasteall.org/pic/show.php?id=21557
> >>
> >> Notes:
> >> The original code was in C++; I converted it to C and switched the
> >> math stuff over to blenlib and the types to MVert/MFace/etc. The
> >> per-vertex output is the two principal curvatures and two direction
> >> vectors. I haven't updated the code to handle quads yet, it'll
> >> assert() if it encounters a non-triangle face. I've only just
> >> gotten
> >> this working, so there are definitely bugs in it yet. :)
> >>
> >> Thanks,
> >> -Nicholas
> >>
> >> [1] http://en.wikipedia.org/wiki/Principal_curvature
> >> [2] http://gfx.cs.princeton.edu/proj/trimesh2/
> >> [3] http://gfx.cs.princeton.edu/pubs/_2004_ECA/
> >> [4]
> >> https://gitorious.org/~nicholasbishop/blenderprojects/nicholasbishop-blender/commits/tri-curvature
> >
> > Hi,
> > There was a gsoc project last year [1] about the implementation of
> > mixed
> > integer quadrangulationsa [2] into blender. I don't know where this
> > project lies now, but I believe this code uses a curvature estimator
> > to
> > generate a direction field on the surface.
> >
> > Thanks for this work,
> > -- Vincent
> >
> > [1] http://code.google.com/p/google-summer-of-code-2010-blender/
> > [2]
> > http://www.graphics.rwth-aachen.de/uploads/media/bommes_zimmer_2009_siggraph_01.pdf
> > _______________________________________________
> > Bf-committers mailing list
> > Bf-committers at blender.org
> > http://lists.blender.org/mailman/listinfo/bf-committers
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers