generate_vertex_normals_ts

[szabolcsdombi]

• [x] Generate normals for indexed triangle strips
• [x] Support primitive restart index
• [ ] Tests for generating normals for triangle strips
• [ ] Docs

[szabolcsdombi]

Unfortunately the winding changes for the triangles generated. Removing the invalid triangles will mess up the winding order.

Example:

index: 1 2 3 4 5 6 -1 7 8 9
triangles:
1 2 3 CCW
2 3 4 CW
4 5 6 CCW
----- primitive restart
7 8 9 CCW

Visual example:

index: 0 1 2 3 4 5 6 7 8
triangles:
(see the image above)

[Korijn]

You'd have to split the array into subarrays first, then call the index restructuring procedure for each of the subarrays, and then concatenate them.

[Korijn]

something like this: https://stackoverflow.com/a/38278327/552379

[szabolcsdombi]

I had no time to push it, but I have some code that can build the indices. I stopped work on this because I noticed the normals are not correct. I will have time around the weekend.

The normals depend on the number of triangles and the "area" of the triangles.

I understand the problem and I want to fix it (maybe it is not a fix but a tweak or a different solution)

I expect to have normals like this:

[Korijn]

You can ignore the area of the triangles to get that result. Maybe add a kwarg weighted=True to the generate_vertex_normals function?

[szabolcsdombi]

This is another example where the normals are not okey. I have an idea but I have not time right now.

[szabolcsdombi]

Yes I will add a kwarg for that.