Author Topic: Seams--Tri Planar Height and Normal Projection Issues  (Read 185 times)

Hey all--methodology question here-
We have a very largescale texturing task at hand--on organic surfaces, with multiple procedurally generated objects that will be calling the same substance designer graph to generate different textures--single varying nodes being position curvature and uv map data per object.   We are trying to everything we can procedurally in designer for the final look, and that includes covering seams.  The solution would ideally be triplanar projection--hammering the seam crevices with grunge or noise that is projected procedurally to just cover our tracks--not just for color but for normal and height maps as well.  Ive noticed though, that triplanar projection has limitations.  If you turn the height displacement up too high--well in certain cases if you turn it on at all, you can see the mesh splitting along the seams despite accurate position and world space maps.  It makes sense--in that triplanar projection isn't just magically gonna make each side of the seam have perfectly lined up values.  Its a very good helper but not perfect.  However the discrepancy in this case is high enough to make height maps a no go along the seams, which is unfortunate.  We can get away with normal maps though, and once the whole object is colored, the seams are very hard to see, but Im just hear to clarify things. 

The only alternative as we see it now is to do this seam edge work on the mesh itself inside Houdini--on a super high res version, and then bake these normal and height maps out in a traditional high res to low res workflow--then we could bring these into substance designer to help leverage the process.  Does that seem like the correct approach to you?  Or perhaps there are some things I am missing. Unfortunately we wont be able to have as much complexity in our seam coverage but it is what it is.   Ive attached a screen shot of a sample object--with the top portion untextured atm.  I do admit we are almost splitting hairs here, but it would be helpful to get and answer on this.  Thank you in advance for your help! 
Last Edit: February 08, 2019, 07:34:53 pm

Does the model have split normals/ hard edges where the seams appear? Those will cause texture seams in triplanar projection and create gaps when using displacement.

Also, have you tried using 3d noises such as 3D worley noise? 3d noise values will line up better than with tri planar, even along hard edges. (though they will not prevent offset splitting along hard edges)
Esger van der Post.
Game design student and texturing addict.

Thanks for the reply--funny, I actually realized that 3d noise could potentially be a solution this weekend--I jumped into houdini to do point based noise deformation on a super high res mesh and realized that 3d noise in sd could achieve the same affect with ease--they both rely on point positions rather than uvs--thats the key.    Along the way it made me realize that I need to do a better job maintaining size between uv shells so there's that lol--you can see resolution discrepancies across the tiles--- though the texture itself no longer has splits.  As for the hardened normals affecting displacement-- Ill have to look into that.  The edges of the object are fused at this point--by that I mean the points are all fused, so the surface is watertight, but the seam edges do have separate vertices because they have separate tiles---does that technically mean they have split normals?  Id imagine not but Ill try to make sure......thanks for the tips!!