Announcement

Collapse
No announcement yet.

Lightcache & Mipmap

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

  • Lightcache & Mipmap

    I saw in the release notes that there were optimizations done, but can I ask which level is selected for lightcache calculations?

  • #2
    The same level that would be selected if a ray was doing mirror reflections instead of GI bounces. The GI rays seem to ignore the VRAY_USE_NORMAL_DIFFERENTIALS environment variable though - this is fixed for revision 14299.

    Best regards,
    Vlado
    I only act like I know everything, Rogers.

    Comment


    • #3
      Nice. This would definitely be a huge performance leap for anyone using GI and MIPMAP exrs.

      Comment


      • #4
        I might have missed that thread... but what does the VRAY_USE_NORMAL_DIFFERENTIALS do? and if we use that i'm assuming we have to add that variable to any render node as well as local machines?

        A

        Comment


        • #5
          Originally posted by mr furious View Post
          but what does the VRAY_USE_NORMAL_DIFFERENTIALS do?
          It causes V-Ray to take into account surface curvature when calculating the ray differentials for reflected rays (the normal mode is to consider the surface as locally flat). This usually translates to lower mip-map levels being needed for textures in reflected objects, while still producing more or less the same visual result. It helps when using tiled OpenEXR files for textures.

          One drawback is that for the moment, the ray differentials computed in this way are not continuous across the triangles of a mesh. In many cases this will not matter, but it is a potential problem.

          and if we use that i'm assuming we have to add that variable to any render node as well as local machines?
          Yep; at least until we decide to include it as a regular render option.

          Best regards,
          Vlado
          Last edited by vlado; 14-03-2011, 01:58 AM.
          I only act like I know everything, Rogers.

          Comment

          Working...
          X