View previous topic :: View next topic |
Author |
Message |
Linker
Joined: 06 Nov 2006 Posts: 29 Location: Almaty, Kazakhstan
|
Posted: Sat Jun 02, 2007 11:40 am Post subject: Ray |
|
|
Hello everyone.
I would like to have my project hosted here.
Ray is a library for your everyday 3D graphics needs.
Regards,
Linker |
|
Back to top |
|
|
h3r3tic
Joined: 30 Mar 2004 Posts: 261 Location: Torun, Poland
|
Posted: Sat Jun 02, 2007 5:32 pm Post subject: |
|
|
Can it do PRT? |
|
Back to top |
|
|
Linker
Joined: 06 Nov 2006 Posts: 29 Location: Almaty, Kazakhstan
|
Posted: Sun Jun 03, 2007 10:23 am Post subject: |
|
|
h3r3tic wrote: | Can it do PRT? |
Nope. Currently, I'm working on mesh simplification and clustering. |
|
Back to top |
|
|
h3r3tic
Joined: 30 Mar 2004 Posts: 261 Location: Torun, Poland
|
Posted: Sun Jun 03, 2007 10:34 am Post subject: |
|
|
Yeey that's cool too Which algorithms? I'm myself looking forward to implementing quadric error metrics. (or letting someone on my team implement them) |
|
Back to top |
|
|
Linker
Joined: 06 Nov 2006 Posts: 29 Location: Almaty, Kazakhstan
|
Posted: Mon Jun 04, 2007 11:50 am Post subject: |
|
|
As of current, it's QEM with half-edge data structure. It's pretty buggy (= only handles 2-manifolds without boundaries), so I'm rewriting it to something more useable. Yes, and it's really slow, because I was lame enough to re-heapify candidates after each decimation step. |
|
Back to top |
|
|
h3r3tic
Joined: 30 Mar 2004 Posts: 261 Location: Torun, Poland
|
Posted: Mon Jun 04, 2007 12:05 pm Post subject: |
|
|
w00t! QEM I wants.
I've also coded some basics of a half-edge data structure. I think it even had the ECOL/VSPLIT operations, but nothing more than that :} |
|
Back to top |
|
|
Linker
Joined: 06 Nov 2006 Posts: 29 Location: Almaty, Kazakhstan
|
Posted: Mon Jun 04, 2007 12:50 pm Post subject: |
|
|
I think that I'll finish rewriting in a couple of weeks. If I have the time, it'll be ready by the end of this week. |
|
Back to top |
|
|
h3r3tic
Joined: 30 Mar 2004 Posts: 261 Location: Torun, Poland
|
Posted: Mon Jun 04, 2007 1:02 pm Post subject: |
|
|
Awesome! Maybe I'll be able to help you out later. |
|
Back to top |
|
|
brad Site Admin
Joined: 22 Feb 2004 Posts: 490 Location: Atlanta, GA USA
|
Posted: Mon Jun 04, 2007 6:57 pm Post subject: |
|
|
Linker,
I suck. Sorry it took so long. Everything is set. Let me know if you have any questions.
BA _________________ I really like the vest! |
|
Back to top |
|
|
h3r3tic
Joined: 30 Mar 2004 Posts: 261 Location: Torun, Poland
|
Posted: Tue Jun 05, 2007 4:14 pm Post subject: |
|
|
Hmmm have you got any plans for handling non manifold meshes in any way? I'm not sure what should be done for them. Should the LoD algo try to cap them by doing some sort of voxelization? Or should just the error metric be modified in a way that allows non manifold topologies? |
|
Back to top |
|
|
Linker
Joined: 06 Nov 2006 Posts: 29 Location: Almaty, Kazakhstan
|
Posted: Thu Jun 07, 2007 12:24 pm Post subject: |
|
|
Big thanks for setting it all up, brad.
Quote: |
Hmmm have you got any plans for handling non manifold meshes in any way? I'm not sure what should be done for them. Should the LoD algo try to cap them by doing some sort of voxelization? Or should just the error metric be modified in a way that allows non manifold topologies? |
Well, it's not the EM but connectivity data structure that's causing trouble; QEM, for example, measures geometrical deviation from original surface (kinda smart words ). And about non-manifolds... Basically, you trade processing speed and simplicity for generality. Garland's scheme does handle non-manifold meshes, by using 'virtual' edges; there're many techniques doing actually the same, but I'm not going to support non-manifold meshes unless anybody can 'convince' me to do so. By the way, the structures I use allow for some degree of 'non-manifoldness'. |
|
Back to top |
|
|
|