[Closed] Getting vertices' position and uv – no correlation?

you can do it this way

The geometry vertices and UV vertices does not necessarily needs to be the same. In fact, most of the times there will be more UV vertices than geometry vertices.

Here are two functions to gather both of them:

(
	fn GetNodeVerts node =
	(
		local getvert = polyop.getvert
		local verts = for j = 1 to node.numverts collect getvert node j
		return verts
	)

	fn GetNodeTVerts node channel =
	(
		local getmapvert = polyop.getmapvert
		local tverts = for j = 1 to (polyop.getnummapverts node channel) collect getmapvert node channel j
		return tverts
	)	
	
	verts = GetNodeVerts $
	tverts = GetNodeTVerts $ 1
	
	format "verts:% tverts:%
" verts.count tverts.count
)

The following simple examples will show how the UV vertices can be completely different than the geometry vertices:

(
	node = converttomesh (box mapcoords:on isselected:on)
	unwrap = unwrap_uvw()
	modpanel.addmodtoselection unwrap ui:on
	unwrap.edit()
	
	messagebox ("NumTVerts: " + (node.numtverts as string))
)

(
	node = converttomesh (box isselected:on)
	uvw = uvwmap maptype:5
	unwrap = unwrap_uvw()
	modpanel.addmodtoselection uvw
	modpanel.addmodtoselection unwrap ui:on
	unwrap.edit()
	
	messagebox ("NumTVerts: " + (node.numtverts as string))
)

PolyTools3D – Thank you very much

I understand now why and how they are not related.

Is there a way to combine them somehow? Without the loop through faces (expensive search for duplicates)?

Only the indices of the mesh faces and the UV faces are the same. The UV verts and edges can be much more than the mesh verts and edges.

I am not sure to understand what you mean by “combine”.

Geometry vertices and UV vertices must be treated separately, as they store completely different information.

Every UV channel is indeed a separated geometry, but we are used to visualize them different than how we visualize objects, and so it might be confusing, but they have their own rules.

I see. About “combining” them: in DirectX I populate vertex buffer with vertices. And each of them has its position (3 floats) and uv coordinates (2 floats). I’ve forgot that 3ds max has a bit more complex interpretation for texture coordinates.

Thank you for clarifications, now I know the “why & how”

Geometry and UV vertices, both have 3 float components, [x,y,z] for geometry and [u,v,w] for UV vertices, which turns out to be pretty much the same as [x,y,z], just different naming.

If you don’t need the W component of the UV vertices (generally not used), you can strip that one out and just have [u,v].

That would give you the same result as what you need to use in DirectX.

[b]

Is there a way to combine them somehow? Without the loop through faces (expensive search for duplicates)?

[/b]yes if you only ever need position and uv coord a simpler method is possible, but as soon as you add normals (every vert can have multiple normals just as every vert can have multiple UV’s) the same with vertex color, vertex illumination, vertex alpha, UV channel 2, UV channel 3 etc etc Each of these can “break” the mesh and in different places and trying to get the simple version to work in these cases can be a tad tricky. IIRC one top studio had a limit on the output number verts as a percentage of the input vert count and if it exceeded it the export would fail so the artist would have to correct the model (stitch up some uv seams for example). If speed is that essential then SDK is the way to go for this sort of thing.

there is an edge thing in mesh geometry (editable mesh, trimesh). there is edge in unwrap modifier.

in map data there are no edges. so it’s hard to say how many of them. one thing i can say for sure that the number of map edges can’t be smaller than the number of not isolated map verts minus 1, and larger than the number of map faces.