of course!OnoSendai wrote:UV coordinates may be a different issue. I guess the question is, will you ever need to have two different uv coordinates at the same vertex position, for different triangles, on a C1 continuous surface?
Indigo 1.1.1
+1. This is bound to happen often, especially in SketchUp.fused wrote:of course!OnoSendai wrote:UV coordinates may be a different issue. I guess the question is, will you ever need to have two different uv coordinates at the same vertex position, for different triangles, on a C1 continuous surface?
Hi F.ip2,
It's hard to say how it affects speed.
The subdivided tris are inserted in the kd-tree like the usual tris, and the kd-tree is pretty fast.
Plus subdivided tris are usually nicely distributed, and nice and small which lends itself to fast kd-trees.
In summary, you'll just have to do a few experiments and see if the speed is acceptable for you
It's hard to say how it affects speed.
The subdivided tris are inserted in the kd-tree like the usual tris, and the kd-tree is pretty fast.
Plus subdivided tris are usually nicely distributed, and nice and small which lends itself to fast kd-trees.
In summary, you'll just have to do a few experiments and see if the speed is acceptable for you
Yeah..we need an answer on this. This has major implications for the exporters. The new features of 1.1.1 are basically unusable in SkIndigo because of this issue.suvakas wrote:Hey Ono,
What are your plans about this dupli verts issue?
Is there any solution you could use or..?
It could be weeks before a new SkIndigo is released if I have to rewrite the mesh exporting code.
Hi,
I'm thinking of a new mesh description format like this:
The changes here are that the uv coordinates are separated from the vertex normals and positions.
Triangles have indices into the vertex pos and normal array, and also into the uv array.
The
line tells Indigo to merge all vertices sharing the same position and normal (uv coordinates will be unaffected tho).
Thoughts?
I'm thinking of a new mesh description format like this:
Code: Select all
<mesh>
<name>mesh2</name>
<normal_smoothing>true</normal_smoothing>
<merge_vertices_with_same_pos_and_normal>true</merge_vertices_with_same_pos_and_normal>
<embedded_2>
<expose_uv_set>
<index>0</index>
<name>albedo</name>
</expose_uv_set>
<vertex pos="-1.5 0 0" normal="0 -1 0" />
<vertex pos="-1.5 0 2" normal="0 -1 0" />
<vertex pos="1.5 0 2" normal="0 -1 0" />
<vertex pos="1.5 0 0" normal="0 -10 0" />
<uvs uv0="0 0" uv1="0 0" />
<uvs uv0="0 1" uv1="0 100" />
<uvs uv0="1 0" uv1="100 0" />
<uvs uv0="1 1" uv1="100 100" />
<triangle_set>
<material_name>mat1</material_name>
<tri v="0 2 1" uv="0 2 1" />
<tri v="0 3 2" uv="0 3 2" />
</triangle_set>
</embedded_2>
</mesh>
Triangles have indices into the vertex pos and normal array, and also into the uv array.
The
Code: Select all
merge_vertices_with_same_pos_and_normal>true</merge_vertices_with_same_pos_and_normal>
Thoughts?
Yes, that's more or less correct CTZn.
The reason i'm using short names here (v, pos etc..), is that when you have large meshes in an .igs file, the amount of characters in the element/attribute names makes a significant difference to file size.
Actually, I think I'll convert more of the names to 1 char names
The reason i'm using short names here (v, pos etc..), is that when you have large meshes in an .igs file, the amount of characters in the element/attribute names makes a significant difference to file size.
Actually, I think I'll convert more of the names to 1 char names
Code: Select all
<v p="-1.5 0 0" n="0 -1 0" />
<t v="0 2 1" uv="0 2 1" />
Who is online
Users browsing this forum: No registered users and 109 guests