Announcement

Collapse
No announcement yet.

Artifacts on certain materials

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Artifacts on certain materials

    Hey,
    I am currently testing the scanned materials. So far so good, but on some objects I get some strange "swirls". See attached pictures. The objects where this material is applied, has proper UV coordinates. The material I used on that object was "leather_a_black_perforated".

    Click image for larger version

Name:	a8lint_vrscans_01.JPG
Views:	1
Size:	172.9 KB
ID:	883829Click image for larger version

Name:	a8lint_vrscans_02.JPG
Views:	1
Size:	209.0 KB
ID:	883830
    https://www.behance.net/Oliver_Kossatz

  • #2
    The scanned materials have a bit stronger requirements for the uv coordinates, testing with simple checker or bitmap is not enough, because the simple maps use just the uv coordinates of the shaded point to calculate their result. Unlike the simple maps, the scanned material uses a property of the uv mapping called "derivative", that gives information how the uv values change in the space near the shaded point. You can test the mapping for defects creating a vray material with anisotropic glossiness, it also uses the mapping derivatives , and shows unexpected hidden defects not visible with simple checker. The most common reason for such defects is hidden small edges or thin triangles, the checker shows perfect squares, but the materials using the uv derivatives are affected significantly. The uv derivatives are interpolation by vray for smooth result and no poly artifacts, but unfortunately gives destructive power to the invisible thin triangles.
    ______________________________________________
    VRScans developer

    Comment


    • #3
      So if I can see these swirls, there is no way to fix it? The geometry is all hand modeled, all quads. No thin polies to be expected there.
      https://www.behance.net/Oliver_Kossatz

      Comment


      • #4
        can you share the scene?
        ______________________________________________
        VRScans developer

        Comment


        • #5
          just fyi , if you have nightly access, there is certain probability to have better results now (i mean with the tomorrows one), we introduced improvements in the algo responsible to catch and hide the bad uv derivatives.
          ______________________________________________
          VRScans developer

          Comment


          • #6
            well, as i expected the problem is related to the uv mapping, they are mirrored in the problematic areas.
            there is a simple workaround, enable the option 2-sided and the rendering will be fine. because many users are using flipped geometry, we will remove the requirement for non mirrored uv mapping, replacing it eventually with a warning in the vray log.
            ______________________________________________
            VRScans developer

            Comment

            Working...
            X