My model has a few vertices that shade abnormally bright, dark, or sharp that I can't easily correct by Shift+N -> Average -> Face/Corner. How can I correct these normals? Here, they're mostly around the lips, though the tragus has something odd too.

This image is updated from this one (which had a more receded browline).

enter image description here

UPDATE: I don't think it's a manifold issue, as the only non-manifold verts are those along the midline of this mirrored half-mesh. The 3D cursor was set to the center of the non-manifold verts; there are no errors in the legs. There are still weird normals near the nose/philtrum and the ear-head junction as above.

enter image description here


1 Answer 1

  1. Fix all non-manifold issues. Use "select non manifold" operation, fix what you see. No edges that connect more or fewer than 2 faces. No zero area faces. No quads with three collinear verts.

  2. Recalculate normals outside, use normal or face orientation overlay to verify is correct (should be, if you've already fixed non-manifold issues.)

  3. Use good topology: all quad, faces as planar as can be, square in aspect and angle, as much as is possible, evenly sized (but all of these constraints will make them evenly sized anyways); no 6+ poles; 3 and 5 poles placed on most planar parts of mesh possible; as few vertices as possible. For deforming meshes, look into typical edge flow with a Google image search.

  4. C-C subdivision, after all other deforming modifiers.

However, here, your mesh has some issues with proportions/placement, and it's easy to get mixed up about what's causing it to not look as good as it could be-- sometimes, it's not issues due to wonky normals, or sometimes, those wonky normals are because the base vert positions aren't great.

  • $\begingroup$ Proportions issues? I'm not going for 100% realism here, but is that really contributing? Since this mesh is a mirrored half, the only manifold verts I got were along the midline. How do I locate edges connecting !+2 faces? $\endgroup$
    – Nintendraw
    Jul 3, 2021 at 23:32
  • $\begingroup$ PS what is the utility of average(normals) -> corner angle (presumably for corner vertices = 3/5-poles? You didn't mention that in your message, but it did seem to fix at least one normal in the nostril. $\endgroup$
    – Nintendraw
    Jul 3, 2021 at 23:36
  • $\begingroup$ @Nintendraw Select non-manifold with "boundaries" and "multiple faces" in operator identifies edges not connecting 2 faces. 3/5, 6+ poles is about the effects of CC subdivision, which doesn't create smooth surfaces except on planar surfaces and 4 poles. What average normals does needs it own question to answer, not gonna happen in comments. Proportion issues are contributing; look at the how the area under your brow is reflecting light that is apparently cast from above. That's because of normals, from proportions-- eyes are too far forward or brow isn't forward enough. $\endgroup$
    – Nathan
    Jul 4, 2021 at 0:38
  • $\begingroup$ I see. When you said proportions, I was thinking "head must be 6-8 eyes wide", etc. I did fix the brow line offline (a bit anyway; time to double-check). Thanks for the clarifications! $\endgroup$
    – Nintendraw
    Jul 4, 2021 at 2:05
  • $\begingroup$ It doesn't seem to be a manifold issue, so time to try the other ones. Not gonna subdivide; I wanna keep polys low. There's gotta be a quick way to locate nonplanar faces, right? $\endgroup$
    – Nintendraw
    Jul 4, 2021 at 2:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.