Baking the appearing seams that Normal cannot fix with Substanc painter in UIDM function
Hi, I'm a scene modeler from China. I'm having some problems baking Normal maps with Sustance Painter.
My model needs two UVs, and the Plane of one of the geometry is distributed in two UVs, I use the UIDM function of Substance painter to import the model and bake it. I found a noticeable seam where the Plane of the two UVs intersect. And when I put this model in the same UV, its baking result is normal, I tried many ways, and searched for this problem and checked the official documentation of Sustance Painter, unfortunately, I did not find a solution way.
I tried simulating the process with a block and found the same result, so, I wonder if this is Sustance Painter's own problem? Or am I doing something wrong?
1.UV distribution
2.High model
3.Import settings and Baking Settings
4.Three faces and their normals in one UV
5.The face and its normal in the three UVs respectively
I checked my smooth group and all possible settings, but it didn't work. Of course, I can use Toolbag or other tools to get the right effect. But now I want to understand why this is, hope some super man can tell me the answer, I will be grateful！thanks.
Replies
Is there enough space between the shells in UV space? Where is the normal-mapped lowpoly model being rendered?
That looks normal to me. The meshes are too different in shape for the distance you are viewing them from.
If you want that to look nice from that distance you need to bevel your low poly
I feel like the point of the OP is being missed here. From what I understand he is stating that given a set of lowpoly and highpoly models, the baking results appear to be different in SP depending on the low having its UVs within a regular 0-1 range, or having them spread as multiple tiles/UDIMs.
If that is indeed the case I'd personally suggest to ignore the SP UDIM feature altogether and instead hard-splitting the model into as many parts as needed (along hard edges of course, so that the re-attaching doesn't cause a change in shading), and reassemblng the UDIM from that later.
Oh. You're totally right..
You can see the normals are coming out differently between the two bakes - it's like the tangents differ when using udim Vs single page
What happens if you bake using the low Res as high or using world space normals?
It smells like a bug tbh - I'd recommend the OP contacts allegorithmic via their forums / support email.
I would check first if uv islands are not rotated differently in single UV vs UDIM. Because shaded screens looks just ok except some tiny thing nobody would care about
That's possible but you'd expect the shaded results to be identical regardless of UV orientation assuming that resolution doesn't change between them
Hi,
Can you disable the padding (dilation width: 0) and see if you have a hit for all pixels?
I suspect the rays don't go far enough and the missing information is filled by the padding, which causes the seams to be visible.
It'd also explain why the rotation produces a different result: because the normal values are different, the padding computation will be different.
If that is the case, simply increase the frontal/rear distances until you no longer see transparent pixels with the padding disabled.