Strange normal map seam.
polycounter lvl 9
Hello,
I am getting a very strange seam right in the middle of an open face. I used symmetry of the clean half to try and fix the issue, I also deleted the half collapsed the symmetry modifier and welded the verts and I'm still getting the issue.
Anyone know what would be causing it?
I am getting a very strange seam right in the middle of an open face. I used symmetry of the clean half to try and fix the issue, I also deleted the half collapsed the symmetry modifier and welded the verts and I'm still getting the issue.
Anyone know what would be causing it?
Replies
Yes, triple checked the UV's, they were all in place. I think that kind of error you are describing happens after the the normal is baked, applied then the symmetry/mirror is done.
@Bardler
Should I be triangulating every time before I bake? :poly122:
But to other applications this means "flip it however they want" which normally defaults to one direction, like this: ////
When the tri-stripping doesn't match, you get seams.
They assume since it isn't a forced edge you don't care and it doesn't matter. Some apps like Maya always force hidden tris to strip one way, which becomes a big pain in the ass when you want something perfectly symmetrical, like joints for example. When everything strips from left to right it doesn't fold the same way. Often mirrored bits in Maya have to be forced to strip another way so they work like they do on the original side, which destroys loops and rings.
3dsmax lets you flip them anyway you want, but then other applications reinterpret max's hidden edges, leading to seams, unless you triangulate.
Really? That's a new one for us I think. Could you please outline your process (what app are you baking in, what file format and settings are you exporting, what tangent space do you selected in TB2, etc). Also a sample mesh+map that shows this issue would be helpful.
If you could post it in the official thread http://www.polycount.com/forum/showthread.php?t=128766&page=21, or just send me a PM that would be great well in case I miss your response here.
1. Triangulate before export, TB2 may triangulate quads in a different direction than max (or any other app), which will result in X shaped smoothing errors
2. For OBJs, Make sure "export normals" is on, "export smoothing groups" is not enough as SGs are a max specific thing that other apps don't really read. Try toggling the optimize settings as well if you still have issues with OBJ.
3. If you're using FBX and getting errors, try OBJ, and vice versa
4. Select the Max tangent space in the per-mesh options (you can set your default tangent space to Max as well and all meshes will come in with max). With the Max TS, you don't need to flip your green as the TS accounts for it.
5.
This is interesting as it's basically the opposite of what I've seen.
In TB1, you would need to offset your mirrored uvs or get strange seams/polygonal lines.
In TB2 this was fixed along with a few other uvs improvements (as far as I know, I don't have the issue on any of my meshes anymore), so you shouldn't need to offset overlapping uvs anymore.
So yeah, if you guys can get me some example content where this happens in TB2 so we can debug it that would be awesome.
I wonder if it has something to do with max's symmetry modifier, have you tried to collapse your stack and reset xform before exporting?
It really depends on what you are doing. Its important to understand the problem first and then you'll be able to answer yourself.
The problem is that various programs may triangulate differently. So when you export your quad mesh to a model viewer, or a game engine, or another 3d app, when it loads your file it the triangulation may change, which will result in smoothing errors.
My basic workflow is usually modeling in Max, baking in xNormal (triangulated meshes via SBM), export identically triangulated mesh (via the Turn To Mesh modifier) as FBX to Toolbag2 and set Tangentspace to xN/Mikk. And yeah those shading edges occur where there's no UV- or Tri edge or anything. Not sure if it's to do with stacked shells, I'll check for that.
Great, yeah if you can provide repro steps and a test asset (even if its a simple cube or something, I understand NDAs etc) that would be very helpful. Thanks!