The dreaded mirrored seam problem
polycounter lvl 4
First time mirroring a character to save texture space and I've come upon the seam that runs down the middle of the character...as well as incorrect lighting on the mirrored side.
Ok, I searched through alot of posts and various sites for hours before I decided to ask here. I went through numerous trial and errors
1. Deleted left half of the mesh, applied the same smoothing groups where there are uv splits/shells. Baked in xnormal with 1-16 of padding. Back to Max, used symmetry modifier...seam remains
2. Used symmetry modifier after making sure each uv split had its own smoothing group. Moved the duplicated UV shells off of the 0,1 space. Bake in xnormal, seam remains.
3. Used symmetry modifier first, then went back and changed the mirrored geomerty to its own smoothing group seperate from those of the right side of the mesh...(ie. left side of face was smoothing group 1, while right side of face is group 2).
4. Used symmetry modifier then applied smoothing groups to both halves so that it 'looked' correct in the Max view port...exported it to the steps above...seam remains...
5. Used Eric Chadwicks photoshop action to add padding the normals as well is case it bleeding occurred.
Im thinking it has to do with the smoothing groups somehow???
I've also baked the maps in topo gun as well, also tried to bake the normal maps using world space, then using Handplane to convert it back to tangent space; through each of the process above...seam remains.
Hope the pictures helps get my point across.
I get the same error in UDK as well.
Thanks for reading...any ideas????
Ok, I searched through alot of posts and various sites for hours before I decided to ask here. I went through numerous trial and errors
1. Deleted left half of the mesh, applied the same smoothing groups where there are uv splits/shells. Baked in xnormal with 1-16 of padding. Back to Max, used symmetry modifier...seam remains
2. Used symmetry modifier after making sure each uv split had its own smoothing group. Moved the duplicated UV shells off of the 0,1 space. Bake in xnormal, seam remains.
3. Used symmetry modifier first, then went back and changed the mirrored geomerty to its own smoothing group seperate from those of the right side of the mesh...(ie. left side of face was smoothing group 1, while right side of face is group 2).
4. Used symmetry modifier then applied smoothing groups to both halves so that it 'looked' correct in the Max view port...exported it to the steps above...seam remains...
5. Used Eric Chadwicks photoshop action to add padding the normals as well is case it bleeding occurred.
Im thinking it has to do with the smoothing groups somehow???
I've also baked the maps in topo gun as well, also tried to bake the normal maps using world space, then using Handplane to convert it back to tangent space; through each of the process above...seam remains.
Hope the pictures helps get my point across.
I get the same error in UDK as well.
Thanks for reading...any ideas????
Replies
http://www.polycount.com/forum/showthread.php?p=1513440#post1513440
@Shadow: I made sure the normals weren't flipped before exporting them...also found an Average normals script on the wiki, no luck...
You might also try exporting the lowpoly mesh from Xnormal as SBM, and importing that into Marmoset. This may transfer the tangents properly.
Method 1 will always create a seam. You shouldn't change the triangles after baking (unless you're using the Handplane method). I don't think Handplane supports Marmoset though, worth asking them.
Also, have you tried inverting the green channel? Have you tried Reset Xform in max before baking?
No guarantees it will work if there are true bake errors, but it's worth a shot.
If you want to use more than one group (i don't think the current mesh needs more than one), make sure they are exported correctly when using .obj.
Thanks alot guys!