Ngon triangulation error #31807

New Issue

Martin Hedin · 2012-06-12T22:51:13+02:00

Martin Hedin commented

2012-06-12 22:51:13 +02:00

%%%When triangulating a cube, with one face sudivided twice, the connections are not as expected. Faces with 0 area are created, and edges of these faces are coincident with the adjacent penpendicular faces.

Blender 2.63 - Ve:8 - Fa:6 - Ob:1-3 - La:1 - Mem:7,45M (0,1m) - Cube
r46461:46487M

Load factory settings.
Select cube -> Edit mode
Sudivide one face twice (W -> Subdivide, 2 times)
Select all faces, triangulate. (A -> CTRL+T)
Pull vertices to reveal 0 area faces.%%%

%%%When triangulating a cube, with one face sudivided twice, the connections are not as expected. Faces with 0 area are created, and edges of these faces are coincident with the adjacent penpendicular faces. Blender 2.63 - Ve:8 - Fa:6 - Ob:1-3 - La:1 - Mem:7,45M (0,1m) - Cube r46461:46487M 1. Load factory settings. 2. Select cube -> Edit mode 3. Sudivide one face twice (W -> Subdivide, 2 times) 4. Select all faces, triangulate. (A -> CTRL+T) 5. Pull vertices to reveal 0 area faces.%%%

Martin Hedin commented

2012-06-12 22:51:13 +02:00

Changed status to: 'Open'

Martin Hedin commented

2012-06-12 22:59:08 +02:00

%%%Would like to comment that this bug makes exporting to game engines, especially Cryengine in my case very difficult, since this error appears very easily, and with conversion to, for example; CGF format, as it can only be done with regular triangulated polygons, and exporter does not accept coincident edges, and faces with 0 area.

(In my case, conversion of DAE to CGF fails, with RC.exe tool supplied with Cryengine3. The DAE contains the errors from the triangulation in the exporting process Blender -> DAE)%%%

%%%Would like to comment that this bug makes exporting to game engines, especially Cryengine in my case very difficult, since this error appears very easily, and with conversion to, for example; CGF format, as it can only be done with regular triangulated polygons, and exporter does not accept coincident edges, and faces with 0 area. (In my case, conversion of DAE to CGF fails, with RC.exe tool supplied with Cryengine3. The DAE contains the errors from the triangulation in the exporting process Blender -> DAE)%%%

Martin Hedin commented

2012-06-12 23:04:37 +02:00

%%%Discussed here: http://blenderartists.org/forum/showthread.php?257705-Ngon-triangulation%%%

Bastien Montagne commented

2012-06-18 19:38:11 +02:00

%%%Here is a patch, which I believe solves the reported problem. Could only test it over a few simple cases, so would be nice if you could stress it with a bit more complex models.

Also, would be nice to compare its quickness with previous code (from official 2.63, e.g.) in a very huge model, didn’t yet made any measurements, new code will be slower but can say how much… :/%%%

%%%Here is a patch, which I believe solves the reported problem. Could only test it over a few simple cases, so would be nice if you could stress it with a bit more complex models. Also, would be nice to compare its quickness with previous code (from official 2.63, e.g.) in a very huge model, didn’t yet made any measurements, new code will be slower but can say how much… :/%%%

Bastien Montagne commented

2012-06-19 19:19:40 +02:00

%%%Just did a small test, and for a 50700 mesh (Suzanne array-duplicated 100 times):

old code: 0.40s
new code: 0.27s

I’m a bit surprised here… but even with ngons object (test cube as described in this report, array-duplicated 500 times, makes 14500 vertices), I have:

old code: 0.13s
new code: 0.09s (more when it has to spit out many printf, as in current patch!).

So it seems performances will definitively not be a problem! :D%%%

%%%Just did a small test, and for a 50700 mesh (Suzanne array-duplicated 100 times): * old code: 0.40s * new code: 0.27s I’m a bit surprised here… but even with ngons object (test cube as described in this report, array-duplicated 500 times, makes 14500 vertices), I have: * old code: 0.13s * new code: 0.09s (more when it has to spit out many printf, as in current patch!). So it seems performances will definitively not be a problem! :D%%%

Bastien Montagne commented

2012-06-22 15:01:52 +02:00

%%%Well, since nobody seems to want/be able to test…

Campbo, assigning to you as you are the ref man for bmesh, here. ;)

The (updated) patch I attached seems to do the job quite well, as far as I have tested it, but I’m absolutely not sure I didn’t miss something important. Anyway, this is a temp fix, as I think there is a GSOC working on a serious triangulation solution?%%%

%%%Well, since nobody seems to want/be able to test… Campbo, assigning to you as you are the ref man for bmesh, here. ;) The (updated) patch I attached seems to do the job quite well, as far as I have tested it, but I’m absolutely not sure I didn’t miss something important. Anyway, this is a temp fix, as I think there is a GSOC working on a serious triangulation solution?%%%

Martin Hedin commented

2012-06-22 18:55:06 +02:00

%%%Sorry,

I am newbie and I am packing for changing country during the summer. Could not test at the moment :(%%%

%%%Sorry, I am newbie and I am packing for changing country during the summer. Could not test at the moment :(%%%

Bastien Montagne commented

2012-06-24 18:24:07 +02:00

%%%fixed in r48243.

Thanks for reporting!%%%

%%%fixed in r48243. Thanks for reporting!%%%

Bastien Montagne commented

2012-06-24 18:24:09 +02:00

Changed status from 'Open' to: 'Resolved'

Bastien Montagne closed this issue

2012-06-24 18:24:09 +02:00

Sign in to join this conversation.

No Label

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Ngon triangulation error #31807