Home Adobe Substance

Normal map shading problem in Substance painter

null
Offline / Send Message
Itam_17 null
Hi, i have problem when i bake normals in substace painter,  the low poly has single smoth group and no hard edges, 
in some in most of the mesh was good even being curved but just in this part flat I get this error. 

Replies

  • NyanBlaster
  • pior
    Offline / Send Message
    pior grand marshal polycounter
    While it may be indicative of an issue, at the end of the day what your model looks like in SP is pretty much 100% irrelevant since SP is obviously not the final environment your model will be displayed in.

    Worry about getting the shading to look perfectly right in your actual target environment first and foremost (UE4 ? Unity ? Sketchfab ?), and then work your way back to SP.

    (BTW this thread should be in Tech Talk, and probably in the Substance subforum)
  • Neox
    Offline / Send Message
    Neox godlike master sticky
    pior said:
    While it may be indicative of an issue, at the end of the day what your model looks like in SP is pretty much 100% irrelevant since SP is obviously not the final environment your model will be displayed in.

    Worry about getting the shading to look perfectly right in your actual target environment first and foremost (UE4 ? Unity ? Sketchfab ?), and then work your way back to SP.

    (BTW this thread should be in Tech Talk, and probably in the Substance subforum)
    With mikkT being the tangentspace in Substance, Unreal and Unity i would assume that the normalmaps should be identical? So issues in Substance should be the same in Engine, given the same geo has been used
  • pior
    Offline / Send Message
    pior grand marshal polycounter
    In theory yes, but in practice nothing prevents human/programmers error, as well as user error too :) Especially with the program being updated quite often, and every step of the pipeline having so many possible export/import settings.

    Back in earlier SP versions the Cyborg Ninja example was completely off in that regard, so even though it is indeed perfectly possible that things are fully synced now (and in my experience they do seem to be), I wouldn't personally rely on it without a proper check in the final environment of choice. Getting that right should always be the top priority imho.

    So at the end of the day, even though the SP display might very well behave as expected ... I'd personally consider it as mostly irrelevant. Still good to have of course.
  • EarthQuake
    Generally this issue shows up when your mesh normals are very extreme. Basically, the resolution of your texture isn't high enough to accurately represent the gradients needed to compensate for the smoothing. If you up the texture res you should notice this getting better.

    Generally what helps in these cases is to use hard edges at your UV borders, which should negate or lessen the problem, though depending on your UV layout this may not help much. You can also cut in extra edges or bevel some of the geometry in this area so the mesh normals are less extreme.

    If this area will be seen up close, adding more geometry may be the best solution. If it will not be seen up close, removing geometry will work too. If you remove the indented circular element you'll reduce the geometry complexity which should sort the problem out.

    This is something that is typical even with a synced pipeline, but Pior's advice to make sure to verify in the final renderer/engine is sound.

    It's an artifact that I generally refer to as "resolution based smoothing errors", you can see some clear examples of it in my big baking tutorial: https://www.marmoset.co/posts/toolbag-baking-tutorial/ - Best Results -> Hard Edges and UVs



    It's not a Substance specific problem, nor is a tangent basis issue, it's a basic limitation of tangent space normal mapping.
  • Itam_17
    Offline / Send Message
    Itam_17 null
    Thanks, did you correct, I read about this tutorial, but there is a handy toolkit for asser, repairs on other bugs, cooked in 2048 anti-x4, the model has no mirrored textures


  • rollin
    Offline / Send Message
    rollin polycounter
    Someone deleted my post  ?!
    Anyhow.. thx eq for the pics.. As I thought.. 
    Hmm.. as I might not change my final texture resolution this needs to be addressed on the model-side
  • EarthQuake
    Itam_17 said:
    Thanks, did you correct, I read about this tutorial, but there is a handy toolkit for asser, repairs on other bugs, cooked in 2048 anti-x4, the model has no mirrored textures


    The problem here is that you've added hard edges where you do not have UV seams, which will cause these errors (also covered in the tutorial I linked earlier). Wherever you have a hard edge you need to break the UVs.
Sign In or Register to comment.