UVDebug: Process BufferGeometry without conversion

[Mugen87]

More Geometry removal...

[Mugen87]

https://rawgit.com/Mugen87/three.js/dev11/examples/#misc_uv_tests

[WestLangley]

if ( geo.isGeometry ) geo = new THREE.BufferGeometry().fromGeometry( geo );

If the user passes in Geometry, won't this generate a plot with face and vertex indices different from the original? I expect that would be quite confusing.

[WestLangley]

Your plots appear to be overprinting triangles on the right edge with two different sets of indices.

[Mugen87]

Your plots appear to be overprinting triangles on the right edge with two different sets of indices.

You're right. I'll have a look.

If the user passes in Geometry, won't this generate a plot with face and vertex indices different from the original? I expect that would be quite confusing.

When converting to BufferGeometry, geometry data become non-indexed so the result will be different. We could stop supporting Geometry to prevent this...

[WestLangley]

We could stop supporting Geometry to prevent this...

I think the only useful solution is to add direct support for both Geometry and BufferGeometry to this utility.

[Mugen87]

Okay, done. UVDebug handles now Geometry and BufferGeometry without conversions.

[WestLangley]

Notice how in the Icosahedron case, the triangles straddle the seam. That is what the special logic was for. :)

[Mugen87]

What do you mean with "triangles straddle the seam"? On what logic do you refer to? Sry, I'm a bit lost^^.

[WestLangley]

For IcosahedronGeometry( size, 1 ), the triangles on the right side of the uv plot straddle the uv seam, that is, they continue onto the left side of the plot. The geometry will only render with a texture correctly if wrapS = THREE.RepeatWrapping. That is what the "// wrap x" inline comments are all about.

[Mugen87]

Um, the visual result is exactly like before...

[Mugen87]

This is how the visualization of IcosahedronGeometry looks in production:

[WestLangley]

I was not being critical. I was just making what I thought was an interesting point. I was reminded of #3299. Sorry I did not make myself more clear.

[WestLangley]

@mrdoob This PR is good-to-go.

[Mugen87]

@WestLangley Thanks for you help in this PR! 👍 😊

[mrdoob]

Thanks guys!