Announcement

Collapse
No announcement yet.

Weird motionblur problem across buckets - no DR

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

  • Weird motionblur problem across buckets - no DR

    Hey

    I have a NDA job due tomorrow which gives me a bit of ache. Have made a crop to illustrate the problem. No spread ofcourse - not that there's much to spread.

    http://henrikbclausen.com/vr_error.mov

    (Edit: Using Vray 2.00.02 on Windows 7 x64)

    Using ADMC 2/8
    Quadratic AA

    Mblur duration: 0,5
    Geom samples: 3
    Subdivs: 16

    DMC Noise threshhold 0,01

    Both my VRay Lights in the scene have 32 subdivs.

    Initially I thought the issues fixed when I changed the Shadow bias of both lights to below 0,01cm - but it still persists on other frames. The problem is not there when Motionblur is disabled.
    Have rendered the whole animation out without mblur now and try to fill in the patches where mblur is needed, but the problem baffles me a bit.

    Any suggestions?
    Last edited by HenrikBC; 18-12-2011, 08:31 AM.
    http://henrikbclausen.com

  • #2
    Fixed AA seems to ..fix.. it. But it is very slow.

    Adaptive subdivision 1/4 also fixes it - but is also slow - and makes the render a bit brighter!?

    I am using Exponential Color Mapping. May this be a suspect?
    http://henrikbclausen.com

    Comment


    • #3
      You could try increasing the motion blur subdivs a bit and see if it helps. If you have subpixel mapping enabled in the color mapping settings, it might also be a reason.

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

      Comment


      • #4
        Yes - I reached the same conclusion with the mblur subdivs a moment ago.
        Is it supposed to work like this? I would expect consistent results - ie. less samples = more noise, and not artifacts appearing below a sweet spot.

        Subpixel mapping is disabled.
        http://henrikbclausen.com

        Comment


        • #5
          Sometimes it happens, yes. When there are too few samples or the noise threshold is too high, V-Ray's adaptive algorithms might cut off sampling sooner than they should leading to slightly incorrect brightness.

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

          Comment


          • #6
            Ok - thanks. Will know where to turn the knobs the next time. Looks like I'll just make the deadline with my tweaks.
            http://henrikbclausen.com

            Comment

            Working...
            X