Substance Painter ver 8.3 bug when baking ID map

[David De Sousa]

When baking ID maps in version 8.3 (Windows 10) there is now a "shadow" of 1 color produced over another color.  See attached picture to see what I mean.  I have not had any probelm baking ID maps from material color in any previous version of Substance Painter.  The issue started for me with version 8.3. 

FYI, my ususal process (which has worked perfectly until version 8.3) is to create a material with different diffuse colors assigned to each material ID for the mesh in 3ds max and then export that color ID mesh via .fbx  and then import into Substance painter as the "high poly mesh" when baking the ID map.   It appears that the new baker engine in 8.3 is creating a shadow (from ambient occlusion  information, maybe?) of 1 color on top of any color that would be below it (i.e., occluded by it).  Obviously, this creates a mess of the color map ID.  This happens for each of 4 different meshes I have tested so far. 

 Is there a fix for this or am I missing some new setting for the new baker in 8.3?  I did watch the video for the new baker window and everything elses works fine for me.  Just the ID map has this bug in it.  Any help would be appreciated. 

[Cyril Dellenbach]

Hi @David De Sousa,

From the picture you sent, I don't think this is shadows, but bleeding. When you bake the model make sure the cage isn't overlapping with other parts of the mesh. You could also make a bake matching by name.

Keep me posted.

[David De Sousa]

Hi Cyril,

Thanks very much for your answer and sorry for the delay in reponding!  You were right about the cage settings.  Cranking back on the max frontal distance removed the "shadows" I was seeing.  However, I still have artifacts in the mesh regardless of the Max frontal distance and max rear distance settings (see attached image of artifacts).  The color map from 3ds max has no artifacts (see attached image of the same mesh with the color map from 3ds max applied).  But when Substance Painter bakes the color ID map artifacts are produced (see attached image of baking settings).   I tried also using the match by name otion you mentioned but it had no effect in this case.  Thanks for your help on this. 

[Cyril Dellenbach]

Hi again @David De Sousa,

On a personal note, I would try to have a cage with a little bit more space, but I am not sure this will solve the issue. Would you agree to share your mesh (or the specific part of your mesh), so I can dig deeper into the problem? You can share it to me privately if you rather not share it to everyone.

Best regards,

[David De Sousa]

Hi Cyril,

I am happy to share my mesh with you.  Please advise where I can send it to your attention.  I assume .fbx format is OK?  Also, I can send you a version with 1 material assigned only ( so that in SP you only get 1 texture set) and another version with color textures assigned by material ID if that is helpful to you.    

Regarding your advice concerning a using a cage with a bit more space, I did try multiple settings with more space on this mesh and it made no difference.  I also tried exporting via .obj instead of .fbx (which I usually use) but the artifacts were still present in the .obj mesh (although they were in different places).   One other thing I noticed is that the artifact seems to be connected with just the ambient occlusion bake in Substance Painter.  If I bake the mesh without Ambient Occlusion then no artifacts are produced.  If I add Ambient Occlusion to the mesh map bakers list (which of course I always want to do), then I get the artifacts.    I do not remember ever seeing these artifacts with previous versions of Substance Painter, but then I have never baked this particular mesh before either.   In 3ds Max, this mesh is well behaved and has no artifacts if I assign materials to it.  Normals are pointing in the right direction for all polygons.  

Thanks for your continued help on this and I will send the mesh when you tell me where to send it.

[Cyril Dellenbach]

Thank you very much. The .fbx format will be perfect. You can send the files to my personal e-mail: cdellenbach@adobe.com

With that being said, if the AO seems linked to the artifacts in your ID Map, I’d advise you to bake first your ID Map without the other maps, and then bake the rest without the ID Map. In the meantime, I’ll investigate to give you a proper answer.

Best regards,

[Cyril Dellenbach]

Hi again @David De Sousa,

After investigation, the issue comes from the UVs. When you bake your ID map, the color from the vertices/materialID/etc. is applied to the map, a map that is limited to a certain number of pixels (1024, 2048, etc.). If the UV islands are too small, there are a lot of chance some artifacts may appear.

Therefore, I'd advise you to work with larger UV islands, a lot of potential issues could be avoided.

Painter auto UV Islands

Small UV islands

Another solution is to increase the output size of your baked maps, but personally this is not the one I would advise you to use.

I hope this was helpfull and have a nice day!

[David De Sousa]

Thanks very much, Cyril.  You nailed it!  I was a bit lazy on doing the UV's on this mesh resorting to brute force flatten mapping in Max without stitching the smaller islands together.   But the reuslts you showed with Substance Painter's Auto UV function look good so I will give that a go next time I'm in a hurry.   The last time I tried the SP Auto UV mapping was whenever it first came out and I can't quite recall why I haven't used it since, but it would potentially be great for speeding my workflow.  Did you just use the default options for the SP Auto UV Unwrap for this mesh?  Thanks again for your prompt and excellent help.

Dave

[Cyril Dellenbach]

The good old flatten mapping! Personally, I now frequently use the auto-unwrap option, and I don't tweak the native parameters that much. Usually, I just check the Avoid elongated UV islands box and I keep the margins small or medium.

Obviously, doing your own UVs will always be interesting, but this is definitely time saving. 

Give it a try!