Does anyone have an example of mirrored UVs working (with no seams) with a normal map in Painter?

--E--

I can’t seem to get mirrored UVs to work 100% with tangent normal maps.  I’ve tried offsetting the mirrored part outside of the 0-1 UV space.  There’s always a slight UV seam no matter what when I crank up the gloss/metalness sliders. 

pior

You left out crucial  factors you can't do without : 

- If you baked using a "half model" with a centerline region that is not perfectly flat across 2 loops, then you'll get an edge no matter what since ray direction will be influenced by only one side of the model. That's why you need to bake from a low that has the full geometry - so that the ray direction evens out on the centerline.

- You also didn't mention if the issue show up in Painter *only*, or if it also shows up in your baking tool of choice and when the model is displayed in its final ingame context.

--E--

Hi Pior, thanks for the reply!

My center line where the UVs are mirrored is not flat.  Maybe that’s my issue?

I’ve only been baking with the full model (with half of the UVs in 0-1 and the other mirrored half in 1-2 UV space) both in Marmoset and in Painter and both show seams when baked. 

I’ve also tried to put those maps onto a full 3D model that has the UVs still mirrored but welded across the seam in 0-1 Space. Still seams. 

Marmoset is my final output context. 

I will reiterate that the UV seam is slight and only really noticeable when the metalness is cranked all the way up. 

Do you have an example of a tangent space normal map working on mirrored UVs with a non flat centerline mesh? Or is that not possible?

pior

Sure yeah, it's definitely possible.

On "My center line where the UVs are mirrored is not flat.  Maybe that’s my issue?" :
Oh nope, I didn't mean to say that it *should* be flat ; I am only saying that one circumstance under which a half-model *could* be used without ill effect for baking is when it is indeed fully flat across 2 loops. But that's almost never the case anyways.

Lastly, two other things that you cannot do without : actually showing pictures of the issue, and uploading files demonstrating the problem. Otherwise the thread will just be conjectures and suggestions - some good, some inevitably bad.

--E--

Thanks again Pior! Here is a download link to a source file test. I've tried these tests several times with different meshes and there's always a seam no matter what > https://www.dropbox.com/s/q9gfo2pdujqb4zz/mirrored_uvs.zip?dl=1

and here are some images of my issue. The seam is way less apparent in marmoset, but still very visible if you look for it.

painter

marmoset

shot of my normals, tangents and uvs on a overlapped and welded version

shot of my normals, tangents and uvs on my separate island version.