CGAL users discussion list

[Rob Patro <>]

Fernando,

I think I may have hacked up some code that works;  however it's
*incredibly slow* because it requires a linear search through the list
of vertices for each vertex split.  Is there a quick way to go from an
id to a handle?  It seems as if the edge_collapse utilized by the
surface simplification package discards/destructs the vertex which is
removed in the halfedge collapse.  However, since the vertex which is
kept in one halfedge collapse may itself be merged in another, I need
some way to go back and forth between a stable identifier (id) and an
unstable one (handle).  This might work if I used a polyhedron based on
a vector and just assured that the discarded vertices aren't actually
deconstructed.  Is this possible?

Thanks,
Rob


Fernando Cacciola wrote:
> Rob,
> 
>> Fernando,
>>
>>   I see what you are suggesting.  Very clever indeed!  My only question
>> is on the ability to *throw away* M1.  Since the decimation works on the
>> actual surface, with vertex and halfedge handles which point to the
>> actual surface, won't you lose M0 during the decimation process?
> 
> Hmm, actually you are correct I'm afraid. There was a bell ringing loud
> as I was writting that but I kept ignoring it ;)
> 
>> Is there a way to run the simplification forward and to collect
> handles (as
>> you did in your pseudo code) which reference the original, as opposed to
>> simplified, mesh?
>>
> I'm afraid this is getting complex :(
> 
> You would have to operate on a copy of M0 to collect the steps, and then
>   "animate" the decimation over the original M0.  Unfortunately, AFAICT,
> the Polyhedron_3 doesn't have a copy operation, so you would have to
> construct two in the same way (say reading an off file twice or so) to
> end up with identical copies.
> 
> Then the steps would have to denote the mesh elements by some stable id
> via some map from id to handle.
> 
> HTH
> 
> Fernando
>