Announcement

Collapse
No announcement yet.

crazy layered opacity slowdown.. impossible render costs. help?

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

  • crazy layered opacity slowdown.. impossible render costs. help?

    I have an earth model im working on. It has several spheres around the earth sphere, one has opacity mapped clouds with displacement or bump on (ive tried both with minimal difference in rendertimes.)

    above that are two spheres with falloff maps and gradient ramps to give the nice blue soft-edged atmosphere.

    rendertimes are awful. If i zoom it till the illuminated part of the earth fills the screen(which i will have to) im looking at 1.5 hr + at 4k, on a 64 core machine. renderfarm costs will be astronomical.

    if i hide the two outer atmosphere spheres, the rendertime for the earth and cloud layer is about 5 minutes at 4k!

    the atmosphere spheres themselves render in about 4 mins.

    what gives?!

    "comp them in post" obviously..

    however if i screen them on in AE, it doesnt look half as nice, i lose the "foggy" effect and i also get the stars behind the atmosphere edge brightened rather than obscured by the atmosphere. im not sure how to comp it in this case.

    basically it would be way easier if i could render in one hit. i dont understand why the rendertimes explode by so much with just one or two extra layers of transparency. any suggestions to speed things up? im already using "stochastic"

    the inner atmospheric sphere i can combine with the cloud layer using blendmtl... this makes a small improvement in rendertime and looks quite similar ( some difference with brightness and bump maps on clouds.. but the other layer needs to be on a larger sphere to get the soft edge to atmos.

    any tricks?

    id send a file but the textures alone are 10+GB






  • #2
    so i simply swapped opacity for refraction with the same map (inverted) and ior of 1.001 and it now renders in 10 mins instead of 1.5 hrs. (happy

    Comment


    • #3
      Sticky!
      Lele
      Trouble Stirrer in RnD @ Chaos
      ----------------------
      emanuele.lecchi@chaos.com

      Disclaimer:
      The views and opinions expressed here are my own and do not represent those of Chaos Group, unless otherwise stated.

      Comment


      • #4
        - i do now have an issue, however, in the alpha channel.. it seems refraction with a falloff gradient in the refraction amount slot doesnt play very well.. at the very outside edge of the atmosphere (which should be totally transparent) i get some glitchy solid bits in the alpha. right along the edges of the outer faces... im going to have to try adding opacity mapping back into the mix with a very tight fresnel to fade out the edge.. hopefully it doesnt kill my rendertimes again.

        Click image for larger version

Name:	alpha issue_diffuse.jpg
Views:	209
Size:	902.8 KB
ID:	1098168
        Click image for larger version

Name:	alpha issue_alpha.jpg
Views:	205
Size:	45.2 KB
ID:	1098169

        Attached Files
        Last edited by super gnu; 08-01-2021, 12:53 AM.

        Comment


        • #5
          Does that go away if you switch from the default microfacet to blinn/phong? Seems the microfacet can mess stuff up in a few situations so maybe worth trying
          https://www.behance.net/bartgelin

          Comment


          • #6
            ive actually resolved it by setting ior to 1 rather than 1.001 or whatever... since i was also getting heavy distortion of satellites passing behind atmosphere edge. which raises the question...

            its kind of a mantra here "dont set ior to 1, always use some number slightly above 1 to avoid problems"

            what problems? its never been clear to me, and in my case seems to fix it. however i dont want any more nasty surprises.

            Comment


            • #7
              No idea where the warning comes from.
              1.0 is a valid IoR, if it isn't found in anything but empty space, in nature.
              Lele
              Trouble Stirrer in RnD @ Chaos
              ----------------------
              emanuele.lecchi@chaos.com

              Disclaimer:
              The views and opinions expressed here are my own and do not represent those of Chaos Group, unless otherwise stated.

              Comment


              • #8
                Cool, glad it's sorted...until the next surprise...there are always more LOL....hopefully not too many though
                tbh I have not seen anything saying to set it to anything but 1 for the equivalent of opacity, so I luckily missed being tricked by that particular oddity.
                It would be interesting to know why that is said, if it does clearly cause issues.

                EDIT: Nice...thanks for clarifying Lele
                https://www.behance.net/bartgelin

                Comment

                Working...
                X