Announcement

Collapse
No announcement yet.

Flame detail scale

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

  • Flame detail scale

    What would be the best method for reducing the scale of flames? I have settings that give me a decent looking flame motion but they are way too big. Here are some settings:
    timescale: 0.5
    Advection Step: 4.0
    Vorticity: 0.5

    Im playing around with advection and vorticity, but not getting where I want.
    Jordan Walsh
    Senior FX artist

    Showreel 2010 (new)
    Check out my scripts at Script Spot!

  • #2
    the scale of the scene is the most important, the time scale gives the same effect. or may be the question is not about this?
    ______________________________________________
    VRScans developer

    Comment


    • #3
      im using the time scale to get the speed correct, but the physical size of the flame detail is too big. I want to reduce the size of the flame detail (the size of the vorticity) by half.
      Jordan Walsh
      Senior FX artist

      Showreel 2010 (new)
      Check out my scripts at Script Spot!

      Comment


      • #4
        i got it, but there small scale details tends to disappear and this is general issue of the grid simulation. you can try the mapping, compose a three component noise texture and set it to attract the velocity.
        ______________________________________________
        VRScans developer

        Comment


        • #5
          Can you explain the 3 component texture in more detail?

          Comment


          • #6
            when the velocity is attracted by map, the r component influences the x component of the velocity etc. set to each rgb component a noise texture, and set to these textures different seeds, to avoid their correlation, this will produce turbulence like force.
            ______________________________________________
            VRScans developer

            Comment


            • #7
              Is this done with Explicit map channels?

              Comment


              • #8
                no, world or object. during the simulation you can't use the uvw channel, that is needed for explicit mapping.
                ______________________________________________
                VRScans developer

                Comment


                • #9
                  Do you use Bercon Noise to get the 3 components?

                  Comment


                  • #10
                    this was just a suggestion, try any type of noise, but do not expect miracles at all. the fine details tend to disappear and this is a fundamental problem of all the grid based solvers.
                    ______________________________________________
                    VRScans developer

                    Comment


                    • #11
                      RGB multiply with the map affecting the alpha seems to work well

                      Comment


                      • #12
                        yes, but this is a render time effect. i think the question was about how to affect the simulation
                        ______________________________________________
                        VRScans developer

                        Comment


                        • #13
                          yes, it is about the sim... I dont think my grid size is small enough to lose the flame detail. What about the other way then. What if i did some flames then the director came to me and said "The flame scale is too small, make the flame scale 2x as big, but dont affect the speed, or the volume they take up" how would you try to achieve that?
                          like the difference between these two photos:

                          Jordan Walsh
                          Senior FX artist

                          Showreel 2010 (new)
                          Check out my scripts at Script Spot!

                          Comment


                          • #14
                            I dont think my grid size is small enough to lose the flame detail.
                            the smallest details are not equal to the cell size, they are bigger with a constant. so if the vorticity and the advection do not give the desired effect, i think the only real solution is to increase the resolution of the grid. even this does not guarantee that the general movement will remain the same, because the vorticity is not scale invariant. my observations are that for keeping the flame behavior you have to increase the vorticity
                            the attached images i made using the same settings, with doubled resolution and doubled vorticity for the second one. the behavior of the fire looks the same, but if only the resolution is doubled the fire does not look the same.
                            Attached Files
                            ______________________________________________
                            VRScans developer

                            Comment

                            Working...
                            X