Announcement

Collapse
No announcement yet.

Best calculation method for texture baking complex geometry

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

  • Best calculation method for texture baking complex geometry

    Dear users,

    Trying out the different calculation methods of Lightcache, Bruteforce, Irradiance Map I have found that the combination of Irradiance Map for the First Bounces and Brute Force for the Secondary Bounces is quite exact and has no false light artifacts like lightcache.

    Does anyone have a hint how to use lightcache in combination with a different method to achieve the same results? Or any other fast method? Maybe I don't know how to tweak parameters for lightcache good enough. Lightcache seems to fail in texturebaking with producing light artifacts very often when baking to texture.


    Maybe anyone can help?

    Robert
    Last edited by Robert1977; 28-04-2008, 01:29 PM.
    Robert

    Max, VRay, Fusion:

    https://www.youtube.com/watch?v=g5fSLrVzpxg
    https://www.youtube.com/watch?v=bpmJgTb_9Ro

  • #2
    Can you post an example of the artifacts that you are getting with the light cache?

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

    Comment


    • #3
      Hi Vlado, thanks for the immediate reply. I have posted two example baked textures for an airgrid. One baked with brute force / lightcache, the other with both brute force for comparison (quite noisy but still exact).

      Thank you very much for your help.

      Robert

      P.S.: Maybe it has something to do with the secondary rays bias? If I switch it to 1.0, the vertical line of white artefacts becomes two lines, but weaker. Don't know really.
      Attached Files
      Last edited by Robert1977; 28-04-2008, 01:52 AM.
      Robert

      Max, VRay, Fusion:

      https://www.youtube.com/watch?v=g5fSLrVzpxg
      https://www.youtube.com/watch?v=bpmJgTb_9Ro

      Comment


      • #4
        Dear Vlado, is this enough info or do you need more?...
        Robert

        Max, VRay, Fusion:

        https://www.youtube.com/watch?v=g5fSLrVzpxg
        https://www.youtube.com/watch?v=bpmJgTb_9Ro

        Comment


        • #5
          How many subdivs do you have for the light cache? Also, what is the actual geometry for this?

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

          Comment


          • #6
            Originally posted by Robert1977 View Post
            Dear users,

            Trying out the different calculation methods of Lightcache, Bruteforce, Irradiance Map I have found that the combination of Irradiance Map for the First Bounces and Brute Force for the Secondary Bounces is quite exact and has no false light artifacts like lightcache.

            Does anyone have a hint how to use lightcache in combination with a different method to achieve the same results? Or any other fast method? Maybe I don't know how to tweak parameters for lightcache good enough. Lightcache seems to fail in texturebaking with producing light artifacts very often when baking to texture.


            Maybe anyone can help?

            Robert
            there are NO rules when doing texture backing, that's my advise...
            Noise is everywhere, bloches, bad colors....
            What you only need is trusting you and you capability with doing things.

            Here is a real ime model radiosity backed by myself (original model from Marko Dabrovic) : http://www.chaosgroup.com/forums/vbu...hlight=sibenik

            Comment


            • #7
              Hi Vlado, I used the following settings:

              Adaptive DMC image sampler:
              Min subdivs: 1
              Max subdivs: 100
              use dmc sampler threshold yes
              lightcache:
              subdivs: 1000
              sample size: 0.02
              scale: screen
              store direct light on
              show calc phase: on
              adaptive tracing: on
              use directions only: off
              number of passes: 4
              filter: nearest, samples: 5

              dmc sampler:
              adaptive amount: 1.0
              noise threshold: 0.005
              min samples: 8
              subdivs multiplier: 1.0

              The real geometry is an airgrid, that is kind of a air inlet with many lamellae, kind of complicated for the lightcache I assume, as I have the same artefacts with rendering the real image. it seems that the lightcache samples are somehow "trapped", not being able to cope with the many inlets of the airgrid.

              I am always trying to minimize artefacts, but what I am trying to do is to find a general method that does without them. Maybe that is not possible.

              @n6: thanks for the tutorial. I trust my cababilities, for sure ;o), but that's not was I was asking. I am familiar with the procedures for texturebaking. Right now its only the "right" calculation method that I am missing. I am using different settings since two years now (irradiance map, lightcache, brute force and the combination of them). But as you said, probably it is always a matter of tweaking and there is no general calculation method. But I have read about the universal vray settings and thought they would be applicable for the texturebaking as well.

              Concerning the tutorial with the cathedral, really nice done. I think today it is possible to display non-square textures on the graphics card, as long as they are of the power of two for example 256 x 128, and maybe this rule is even no longer valid. You can crop the baked texture and the uv-map after texturebaking with a custom maxscript so that a rectangular lightmap would be optimised, if you need to texture bake a long thin object into square.
              Last edited by Robert1977; 29-04-2008, 01:55 AM.
              Robert

              Max, VRay, Fusion:

              https://www.youtube.com/watch?v=g5fSLrVzpxg
              https://www.youtube.com/watch?v=bpmJgTb_9Ro

              Comment

              Working...