EDIT: Although it would feel more tidy if FilletEdge produced a UV aligned fillet when both input UVs are aligned, this has been transformed into a wish for the anisotropy option of a PBR material to have texture mapping channel option.
ORIGINAL:
When both input surfaces have aligned UVs, fillet edge produces a surface that opposes both inputs.
(In this video the green isocurves represent the U direction) FilletEdgeUVs.3dm (96.2 KB)
@pascal It’s the same with v5 … so it’s been like this for a while. Is there a specific reason why it doesn’t pay attention to the UVs ? Is it something that can improve ?
Possibly most people missed it, because it’s inconsequential a lot of the time ? or they don’t use unique UV display color to notice ?.. but if you apply a render material to it with anisotropy enabled then all corners look wonky and you have to go into the model to either fix all these fillets or start applying manual UVs in bunch of places.
Hi Thomas - off hand I doubt that it can be ‘fixed’; it simply is not considered and never has been. In any case, it would only be non-sloppy in a minimum set of cases.
@pascal There is NO texture. We don’t want to texture every nook and cranny of a model. A lot of objects in a scene, especially bare metal are just a metal material with some anisotropy.
Yes, indeed, sorry, I missed that bit.
Here, anisotropy looks better if I apply box mapping to the objects. This can be done to multiple objects in one go.
… Unless it’s cheaper for the anisotropy in Cycles to have a WCS option next to it (without having to apply a texture) ? Essentially, anisotropy is a form of a texture, since it gives the semblance of toolmark direction.
hmm, never mind, even if we did apply textures, I notice the box WCS option just shifts the problem, now the discontinuity is at the middle of the fillet.
hmm (thinking out loud) … even if FilletEdge paid attention to the UVs at position 1 (rule: obey the UVs when both inputs are consistent), then a fillet at the top would be OK at position 2, but have a problem at position 3 … unless there is a complex set of rules.
Anisotropy creates a set of problems. We can’t use it willy-nilly.
Right, matching up UVs is only going to help, at best, in a very few cases. I do not know how anisotropy is mapped, but @nathanletwory will have a much better idea than I do…
Nathan, Thomas is running into problems where either seams between surfaces or edges of the box mapping show up in anisotropy in PBRs… I guess there is not a way around that currently, but you know best…?