Author Topic: Seam depending on UV island rotation (viewport issue)  (Read 5069 times)

In Painter i have an issue in the 3d viewport where there is an UV seam/cut:
if two adiacent edges don't have the same "orientation" in the UV layout then the faces normals displayed on the SP viewport are not "correct".
However seems like is just a visual issue with the SP viewport as the normal map is exported correctly, in the example below as flat values.

Hopefully the example image and fbx can explain better.
First square (from left) don't have problems. The others with UV islands rotated show the problem, they give the impression the faces are tilted even if the mesh is flat and there is no normalmap.

The problem is from my side or from SP? Any way to fix the issue, apart from rotating the UV islands edges (which i guess would be nearly impossible on more complex meshes)?

edit: the material applied to the planes is a fill with uniform base color, roughness at 0, no normalmap/bump or additional maps.
Last Edit: January 16, 2016, 02:02:15 pm

are you trying to get the pattern to go the same direction?  If so, turn on tri planar projection on the fill layer, you can see in my example that I have attached.  But you should try and get the UV to go in the same direction if possible.

-G

are you trying to get the pattern to go the same direction?

-G

Thanks for checking but no, in my example the material applied to all three planes is simply a very dark plain base color + roughness set at 0 to show better the problem. Those seams should not be there unless i'm overlooking something...

Hey mik1190,

It seems that you're rotating the UVs but not the material with them. this is obvious when you look at your pictures.

You need to either use a tri planar projection to 'tell' Substance Painter to take the actual 3D mesh into account or rotate one half of the layer manually.


Best regards
Environment Artist - Twitter

Fabian: i forgot to specify in the first post that the material applied to the planes is "basic", triplanar seems like is not solving the issue.

I have attached the .spp file, zipped, should explain even better :)

Greetings

What's your normal map like ?
How did you compute the tangent basis ?

Cyrille: there are no normalmaps and bumps...
I'm not sure about second question. Is the option in SP when importing the mesh? Tried with checkbox checked and uncheked but result is the same.
The faces slightly change "tilt angle" only when switching Nmap format from one to the other (even if there is no Nmap imported/used in the project).

What file format did you use to import your mesh ?
An FBX can contains Tangent and Binormals, but not OBJ. That's what Cyrille si asking, but the Tag/Bin have been generated in your application or via Painter.
Don't forget your log file. It can be exported from the Help menu of the software.
-----------
Fabrice Piquet aka Froyok. Product Manager, Technical Artist and Documentation at Adobe.

What file format did you use to import your mesh ?

I used the .fbx  attached in post #1, exported from Blender, but you can directly test the .spp in post #4 .
Seems like there is no difference between obj and fbx, even with the option to include binormal and tangent vectors.
Out of curiosity, am i the only viewing those seams?

We reproduced the problem on our side and identified two reasons as for why :
- The default for a normal is 127, 127, 255 (which should be 127.5, 127.5, 255 but that's not possible in a 8bit format).
- The specular and diffuse reflection works by using samples to read the env-map, the less you have them, the more divergence will appear.

The first point will be solved on our side, the second one can only be "solved" by increasing the amount of samples (see the quality setting in the shader parameter). However unless you increase it to a very crazy number (which is really not a good idea if you want good performances) some differences will still be visible.
Don't forget your log file. It can be exported from the Help menu of the software.
-----------
Fabrice Piquet aka Froyok. Product Manager, Technical Artist and Documentation at Adobe.

The first point will be solved on our side, the second one can only be "solved" by increasing the amount of samples

Good to know that, thanks!