Rendering points and line segments

[peteroupc]

The glTF 2.0 specification should include advice on rendering points and line segments (primitives with mode LINES, LINE_STRIP, LINE_LOOP, or POINTS), especially because the lighting calculations rely on vertex normals, which don't normally make much sense for line segments and points to have because they have no area. The specification should especially mention how the base color, emission, per-vertex colors, and other parameters affect how line segments and points are rendered.

[pjcozzi]

Thanks for comment!

Discussed offline:

• if normals are defined, use full PBR model
• if normals are not defined, use full PBR model with the assumption that the line is lit from all directions

@emackey has volunteered to make the changed

[lexaknyazev]

@emackey While making those changes, please review existing impl. notes on attributes and lighting calculations (right before morph targets).

Also, spec should be more clear about:

• Does TANGENT attribute always require NORMAL? (I assume yes.)
• Is it OK to have POINTS with TANGENT?

[emackey]

Do we have any sample models that show points and lines?

[pjcozzi]

@emackey did you make these spec updates?

[lexaknyazev]

@pjcozzi @emackey
Any actions left?

[emackey]

I didn't make any updates myself, but I recall there were changes at some point to the relevant part of the spec for unrelated reasons. This is what it looks like now:

https://github.com/KhronosGroup/glTF/tree/master/specification/2.0#point-and-line-materials

If that's OK with everyone, I vote we close this.

[peteroupc]

I didn't make any updates myself, but I recall there were changes at some point to the relevant part of the spec for unrelated reasons. This is what it looks like now:

https://github.com/KhronosGroup/glTF/tree/master/specification/2.0#point-and-line-materials

If that's OK with everyone, I vote we close this.

I have no objections with what that section currently says. Thus, this issue is closed now.