Announcement

Collapse
No announcement yet.

Glossy surfaces and slow renderings

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

  • Glossy surfaces and slow renderings

    I have an image that has stainless steel on many surfaces. Infact a whole wall is SS tile. I have found if I set the subdivions really high on the material up to 70 I have a pretty clear image. But the cost is a 50 hour rendering. I have played with irr map settings and it seems that the only control I have over the noise is subdivions. Is there anything else that will clean this up? I am using lc for secondary and I do have the use glossy on.

    I will post an image to illustrate

  • #2
    funny we've been talking about similar problem on another forum. Try setting the glossy subdivisions to 8, and adjusting the qmc sampler instead. Try higher sampling such as 1/64 - 8/64 with contrast of 0.01-0.001.
    Dmitry Vinnik
    Silhouette Images Inc.
    ShowReel:
    https://www.youtube.com/watch?v=qxSJlvSwAhA
    https://www.linkedin.com/in/dmitry-v...-identity-name

    Comment


    • #3
      Thanks Dmitry - I am rendering a region that is how I am testing my settings at full res 3000 pixels wide. Unfortunatly when I increase the QMC settings the sample rendering time went from 15 min to over 1 hour. I am asuming you meant the image sampler. Right?

      Comment


      • #4
        "contrast" - ur talking like a MR user
        I just can't seem to trust myself
        So what chance does that leave, for anyone else?
        ---------------------------------------------------------
        CG Artist

        Comment


        • #5
          I would suggest you to evaluate what sampler is doing what:
          Try rendering the glossy effects to their default value, with Fixed AA.
          Then slowly raise the materials subdivs until you think the AA is going to do a lot LESS work (ie. the glossy looks "allright" even without AA; say up to 32 subdivs for a 0.6 glossy).
          Then the QMC AA will just be there to help clean, not to actually calculate the glossies, as in the Universal (or semi-universal) Method.

          Reducing noise threshold/adaptive amount helps, but first make sure your issue is about reaching the Max subdivs in either the material or the AA (rendertimes will tell you), before actually lowering those, as they influence a lot of other QMC calculations (GI for instance).

          Also, stop looking at noise at 400% zoom!

          Lele

          Comment


          • #6
            Originally posted by Paul Oblomov
            "contrast" - ur talking like a MR user
            you know I didnt even realize I wrote that but yeah Im now semi mr semi vray user...just got to do that...since vray for maya has ways to go yet.
            Dmitry Vinnik
            Silhouette Images Inc.
            ShowReel:
            https://www.youtube.com/watch?v=qxSJlvSwAhA
            https://www.linkedin.com/in/dmitry-v...-identity-name

            Comment


            • #7
              I am working on it again. Thanks Lele for the tips.

              Comment


              • #8
                I am getting the time down but it is still in the 40 hour range and that is just too long. Any other suggestions? Is there a way to fake glossies?

                Comment


                • #9
                  You might try Ward shader + High Anisotropy (.7 - .9)+ slight glossie (.8-.9) reflections and an appropriate diffuse and bump maps for the stainless.
                  Eric Boer
                  Dev

                  Comment


                  • #10
                    O k I will try that approach as well. Thanks Rerender.

                    Comment


                    • #11
                      I can say that PPT (biased mode, non adaptive) helped me when i had an elevator cabin to render, which had glossy interreflections.
                      Compared to the standard approach, it was WAY faster, and a lot less sensitive to glossies, as such.
                      I'm suggesting this given the amount of time your image is rendering for...
                      But maybe it's just a 15 meters poster, so i may be wrong :P

                      Lele

                      Comment

                      Working...
                      X