[render] RenderMaterial offers more nuanced and complete warnings

[SeanCurtis-TRI]

The fallback materials would use the ('phong', 'diffuse_map') or matching foo.png images as diffuse texture map, even if the underlying mesh could not support it. This is because they were given no information about whether a texture could be applied or not.

Now, it is provided information as to the support of texture coordinates and will also omit a texture if the underlying geometry lacks sufficient texture coordinate support.

Note: This leads to a very subtle, but non-destructive behavior change. If a diffuse map was specified for a capsule in RenderEngineVtk, the image will not appear (as it did before), but now there's a warning emitted.

This is the first PR in a train relating to #17266. The full PR train can be seen in #20235. This PR corresponds to the first commit in that PR.

---

This change is 

[SeanCurtis-TRI]

+@zachfang for feature review, please.

Reviewable status: LGTM missing from assignee zachfang, needs platform reviewer assigned, needs at least two assigned reviewers, missing label for release notes (waiting on @SeanCurtis-TRI)

[zachfang]

feature.

Reviewed 11 of 11 files at r1.
Reviewable status: 2 unresolved discussions, needs platform reviewer assigned, needs at least two assigned reviewers, missing label for release notes (waiting on @SeanCurtis-TRI)

---

geometry/render/render_mesh.h line 33 at r1 (raw file):

  /* Indicates the degree that UV coordinates have been assigned to the mesh.
   only UvState::kFull supports texture images. No matter what, the `uvs` matrix

nit.

Suggestion:

Only

---

geometry/render/render_mesh.cc line 273 at r1 (raw file):

  // If we've only used a single material in the OBJ, use it. If the only
  // referenced material is "-1", it means there is no material.
  mesh_data.uv_state = has_tex_coord ? UvState::kFull : UvState::kNone;

BTW/question. To confirm, has_tex_coord is a hit-or-miss from the current tinyobjloader parsing. Thus kPartial is not used other than the specific test cases we set up.

Code quote:

mesh_data.uv_state = has_tex_coord ? UvState::kFull : UvState::kNone;

[SeanCurtis-TRI]

+@ggould-tri for platform review, please.

Reviewable status: LGTM missing from assignee ggould-tri(platform), missing label for release notes (waiting on @SeanCurtis-TRI)

---

geometry/render/render_mesh.cc line 273 at r1 (raw file):

Previously, zachfang wrote…

BTW/question. To confirm, has_tex_coord is a hit-or-miss from the current tinyobjloader parsing. Thus kPartial is not used other than the specific test cases we set up.

The has_tex_coord variable disappears in the next PR, replaced by meaningful computation of the uv state. At that point, we will be able to meaningfully indicate whether they have no UVs or just partial UVs.

[zachfang]

Reviewed 2 of 2 files at r2.
Reviewable status: LGTM missing from assignee ggould-tri(platform), missing label for release notes (waiting on @SeanCurtis-TRI)

[ggould-tri]

as written but I have one design query below.

Reviewed 11 of 11 files at r1.
Reviewable status: 1 unresolved discussion, missing label for release notes (waiting on @SeanCurtis-TRI)

---

geometry/render/render_material.h line 15 at r1 (raw file):

/* Reports how UVs have been assigned to the mesh receiving a material. Textures
 should only be applied to meshes with *fully* assigned UVs. */
enum class UvState { kNone, kFull, kPartial };

minor: I don't understand how kPartial comes about in our code. I see that kFull and kNone both occur in certain default conditions, but I don't see how we ever get kPartial.

[SeanCurtis-TRI]

Reviewable status: missing label for release notes (waiting on @SeanCurtis-TRI)

---

geometry/render/render_material.h line 15 at r1 (raw file):

Previously, ggould-tri wrote…

minor: I don't understand how kPartial comes about in our code. I see that kFull and kNone both occur in certain default conditions, but I don't see how we ever get kPartial.

Zach had a similar question. It becomes meaningfully populated in the follow-up PR (the second commit in the linked epic PR).

[ggould-tri]

Reviewed 2 of 2 files at r2.
Reviewable status: missing label for release notes (waiting on @SeanCurtis-TRI)

---

geometry/render/render_material.h line 15 at r1 (raw file):

Previously, SeanCurtis-TRI (Sean Curtis) wrote…

Zach had a similar question. It becomes meaningfully populated in the follow-up PR (the second commit in the linked epic PR).

:) In the future perhaps put in a temporary TODO comment in such cases to spare yourself two needlessly confused reviewers.

[ggould-tri]

Reviewed all commit messages.
Reviewable status: missing label for release notes (waiting on @SeanCurtis-TRI)

[SeanCurtis-TRI]

Reviewable status: missing label for release notes (waiting on @SeanCurtis-TRI)

---

geometry/render/render_material.h line 15 at r1 (raw file):

Previously, ggould-tri wrote…

:) In the future perhaps put in a temporary TODO comment in such cases to spare yourself two needlessly confused reviewers.

Touche

[SeanCurtis-TRI]

+(release notes: none)

Reviewable status: complete! all discussions resolved, LGTM from assignees ggould-tri(platform),zachfang