Is this a bug or the mesh itself?

[Operatik]

I found a weird coloring bug on this mesh part, not mine, but I got permission for it.

Reproduction is easy:
Step 1: Find a meshpart.
Step 2: Color it to blend other parts in workspace.
Step 3: Rotate the mesh randomly, until you see the defection.

Other Information:

• Lighting is the source of the problem, it causes weird coloring on daytime, shifting it to nighttime causes no problems.
• This occurs only in a specific rotation of the mesh.

Question is if it is the mesh itself or an actual bug of the client.

[Veldaren]

Not enough context. What am I looking at?

[Operatik]

Forgot to add up this detail: Both the mesh part and the parts surrounding it has exactly the same Color3. However, viewing upon the parts, they don’t match up.

[Veldaren]

Are other properties all the same?
Does the color match if you make the Parts into UnionOperations?

[Operatik]

I’ll let you know tomorrow, it’s approximately 12 AM here. Yes, their properties are the same, but the only factor that touches it is rotation/orientation and the lighting.

lol i don’t want any friggin unions with the mesh parts, I’ll get a better view of it soon or even a file demonstrating it

[Veldaren]

Yea, I just want to know if Unions will match MeshPart’s color. There was some issue related to that, and it was caused by the Show Outlines property under Workspace (IIRC).

[Operatik]

Well then, this will probably blow up the entire building operation, unless I trick with the lighting.

The mesh part has in fact a color that should match 97% of the building. The mesh is a curved inner corner.

[AllYourBlox]

Is the mesh part non-uniformly scaled, by any chance? I’m just wondering if it’s possibly related to non-uniformly-scaled meshes not always having correct surface normals.

[Operatik]

That gotta be it, the size of the mesh is Vector3.new(13, 19, 17)… and I’m quite confused what is counted toward non-uniformly scaled mesh parts.

---

Screenshots Test

Screenshot A: Stretched mesh part, bugged rotation, random size

Screenshot B: Squared mesh part, bugged rotation, 8x8x8

Screenshot C: Squared mesh part, bugged rotation, 2x2x2

Screenshot D: Squared mesh part, rotated once from original orientation, 2x2x2

Screenshot E: Stretched mesh part, rotated once from original orientation, random size

---

Conclusion

It is difficult to tell if the size matters, but there is a noticeable difference between larger mesh parts from smaller ones and also different rotation.

Can I get a background information of what’s going on here?

---

Edit: As a side note, the parts are resized evenly with whole studs, no decimals left. I also made a final test:

Screenshot of A, but with double digit decimals on each axis.