Author Topic: Seams at normals with triplanar projection fill layer [SOLVED]  (Read 3076 times)


I must be doing something wrong with Substance Painter but I can't figure out what's the problem.

Basically I import a mesh, and use a custom material as a fill layer with triplanar projection. All maps generated are seamless except the normal map. However the normals generated by the custom material itself doesn't seem to have seams.

I'm kind of lost... Any help would be appreciated.

If I change the channel mapping of the diffuse to show the normal map, I get no seams when viewing the normal as diffuse:

Viewing the normal show seams, map layout changes:

Thank you very much!
Last Edit: June 29, 2018, 03:00:59 pm

Forget it... I think I was too focused on that normal view.
When on the material view there's no seam at all, I've checked it disabling the diffuse and it's all good.

Nope I wasn't wrong...
I really have a problem with that normal seams.

I've tried to repeat the process on a simple torus, I have noticeable seams under Substance Painter and they are very noticeable when it's exported to Unity.

My process is the following:
- I create a torus in Blender, set smooth faces, and auto smooth to 60°, I make my seams and unwrap then I put a triangulate modifier. I exported in FBX format, made sure that Apply Modifiers was ticked.
- I import it in Substance Painter and apply a fill layer with triplanar projection.

What I'm doing wrong ?

This is normal, you're not doing anything wrong per say.
In real time 3D, UV seams actually split the vertices in 2 along that seam.
So where you had 1 vertex, you now have 2 vertices with slightly different normals, hence the seam in the normals. The only way to solve this issue it to bake a normal map from a high poly torus/rock, and apply your material on top of it.

Thank you for the quick answer!

So that split tangents makes the normals ponting in different direction at split vertices I guess ?

I had a hi-poly for the rock but I still had seams, so I suspect my process of authoring the hi poly is wrong and I have a few questions to ask:
- should the high poly completely wrap the low poly model, or it isn't really important ?
- do I really need a cage made from the low poly that wraps both hi-poly and low-poly or it isn't really necessary for good result ?
- if I take my low poly torus and subdivide/smooth it, is it sufficient to consider it as a workable hi-poly model ?
- should the hi poly and cage model be triangulated ?

EDIT: I made a hi-poly torus by subdividing and smoothing it, then baked the maps in Substance Painter giving the hi-poly model in baking parameters but I still have normal seams...
Last Edit: June 28, 2018, 03:47:44 pm

Solved everything using hi-poly model as well as a cage. Thank you very much for your help!