Faces are lost after triangulation #32304

New Issue

Yousry Abdallah · 2012-08-10T16:42:40+02:00

Yousry Abdallah commented

2012-08-10 16:42:40 +02:00

%%%After triangulating (CTRL-T) a model with ngons some faces are lost.%%%

Yousry Abdallah commented

2012-08-10 16:42:40 +02:00

Changed status to: 'Open'

Sergey Sharybin commented

2012-08-10 16:52:32 +02:00

%%%Seems that triangulation operator is using different approach than triangulation used for display. Bastien, could you please check this?%%%

Bastien Montagne commented

2012-08-13 21:44:05 +02:00

%%%Imho, this is not a bug. We know we have two different triangulation algorithms (one RT, and one for the triangulation op), and that neither of those are really great… Iirc, there was a GSOC project on this topic, but didn’t heard about it. so I guess this remains a TODO.%%%

Yousry Abdallah commented

2012-08-14 09:37:55 +02:00

%%%Sorry, but why do you think that an essential mesh transformation that is now unusable is not a bug?

Is there a workaround available?

I'm using blender to create models for my OpenGL applications. Without a sound model triangulation I have a lot of extra work to clean up the meshes.

Why is it so complicated to exchange one triangulation-algorithm with another? Perhaps I can help. Could you show me the entry points in the sources?%%%

%%%Sorry, but why do you think that an essential mesh transformation that is now unusable is not a bug? Is there a workaround available? I'm using blender to create models for my OpenGL applications. Without a sound model triangulation I have a lot of extra work to clean up the meshes. Why is it so complicated to exchange one triangulation-algorithm with another? Perhaps I can help. Could you show me the entry points in the sources?%%%

Bastien Montagne commented

2012-08-14 10:43:40 +02:00

%%%How do you say “unusable”? If it is because it generates another triangulation than the RT algo over a non-planar ngon, you can’t call it “unusable”! But the two pictures you gave make it hard to realize what exactly happens, I’d be glad to check your Blender file (it’s always a good idea to upload a blend with a simple case of the bug, it helps us much!).

Note however that my “blender” time is very limited these weeks, so I may not be highly responsive…%%%

%%%How do you say “unusable”? If it is because it generates another triangulation than the RT algo over a non-planar ngon, you can’t call it “unusable”! But the two pictures you gave make it hard to realize what exactly happens, I’d be glad to check your Blender file (it’s always a good idea to upload a blend with a simple case of the bug, it helps us much!). Note however that my “blender” time is very limited these weeks, so I may not be highly responsive…%%%

Yousry Abdallah commented

2012-08-14 13:01:28 +02:00

%%%Creating non-planar 4+ nGons is almost unavoidable during the modeling process. I created a demo .blend file but I think you already understand the problem.

This is especially ugly if the offset is minimal. For example due to rounding errors.

Also during the triangulation, sometimes malformed and useless triangles are generated (sometimes even with counter clockwise direction).

I just realized that the original nGon is already destroyed if you add a texture. I think you don't have two but at least three different triangulation algorithms ?!

I added this problem also to the demo file.

Do you know the delaunay triangulation algorithm? Do you always check the network topology?

BTW: My time is also very limited. I just try to help making Blender an alternative to professional software.
%%%

%%%Creating non-planar 4+ nGons is almost unavoidable during the modeling process. I created a demo .blend file but I think you already understand the problem. This is especially ugly if the offset is minimal. For example due to rounding errors. Also during the triangulation, sometimes malformed and useless triangles are generated (sometimes even with counter clockwise direction). I just realized that the original nGon is already destroyed if you add a texture. I think you don't have two but at least three different triangulation algorithms ?! I added this problem also to the demo file. Do you know the delaunay triangulation algorithm? Do you always check the network topology? BTW: My time is also very limited. I just try to help making Blender an alternative to professional software. %%%

Yousry Abdallah commented

2012-08-14 13:05:48 +02:00

%%%Demo .blend added%%%

Bastien Montagne commented

2012-08-14 22:43:56 +02:00

%%%This is definitively not a bug. A bug is something not working as intended, not something working not as good as you might expect, but doing the job anyway! Current tri algo works, in the sense it does not generates double faces, crossed faces, holes, etc.

Again, we do know neither the RT nor the op algo work ideally, this was one of the GSOC that failed this summer, so it’s back on our todo list… This spring I just made a completely broken code working again, roughly reusing what existed for old quads, but as the GSOC was about to start, didn’t spend too much time on it…%%%

%%%This is definitively not a bug. A bug is something not working as intended, not something working not as good as you might expect, but doing the job anyway! Current tri algo works, in the sense it does not generates double faces, crossed faces, holes, etc. Again, we do know neither the RT nor the op algo work ideally, this was one of the GSOC that failed this summer, so it’s back on our todo list… This spring I just made a completely broken code working again, roughly reusing what existed for old quads, but as the GSOC was about to start, didn’t spend too much time on it…%%%

Brecht Van Lommel commented

2012-08-30 19:15:30 +02:00

Changed status from 'Open' to: 'Archived'

Brecht Van Lommel closed this issue

2012-08-30 19:15:30 +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

Faces are lost after triangulation #32304