Announcement

Collapse
No announcement yet.

Particle Shader Wishlist

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

  • Particle Shader Wishlist

    Here are some things that would make us more productive with Particle Shader.

    * Ability to AND subgroups together, like Smaller than 0.1 AND Above water

    * Ability to make subgroups based on particle RGB color (or otherwise associate them with a particular PHX Source emitter vs. the foam generated "on hit," etc.)

    * Ability to use Cutter Geometry with Geometry Mode

    * Ability to export mattes like Multi-matte in Volumetric mode.

    * Ability to use custom geometry (We can use Instancer or Frost, but then we lose some of the other subgroup controls, etc.)

    * Ability to at least randomize (control ably) the squash and stretch of the bubbles to give them at least a little variance. Include stretch along motion vector.

    * Might there be some way to offer some blur (literally softening) of the bubbles, etc. Maybe this is high-frequency motion noise with motion blur, but some way to take the edge off the bubbles without having to blur them in post. Sometimes motion blur isn't enough to do this, or certain particles don't have enough motion blur, or it is in the wrong direction. Some way to have each particle get some random sub-frame motion (jitter) to get some softening via motion blur would be really helpful and avoid hand painting of blurring in the post. Sort of a Heisenberg Principle of particle positions except we don't really knows its velocity or its position. We would of course want to be able to adjust the strength of this jitter. I guess this would require multiple geometry samples, and maybe get crazy slow, but perhaps there is some trick. Some camera space convolution or something. Or perhaps animated scaling. Maybe this is even possible now with sub-frame animated scale noise. Hmmm.

    * Perhaps the above is a fresnel-like (towards-away) transparency at the edge of the bubbles-- a way to soften the edges, We would have to adjust the highlights separately.

    Those are some of things I have been wanting lately. Thanks again for listening.

  • #2
    Oh yes, I already have in mind a solution for (1) and (2). As for Geometry mode, it might be much more beneficial to make render elements work without it...

    The custom geometry and non-spherical shapes would be a job for a different node - there is a cluster of similar requests that I have to generalize though, so not sure yet what the final thing might be, but please keep em coming, this is very useful info!

    Indeed the naive solution to blurring the particles by jittering like with motion blur or dof would make rendering terribly slow. Note that Point mode kinda gets the blurred look with a very large point size automatically. However, there all particles are in camera space, so you wouldn't get the ones near the camera zoomed in and the ones furhter away zoomed out, which limits the application of large points...
    Svetlin Nikolov, Ex Phoenix team lead

    Comment


    • #3
      Cool on 1 and 2.

      Yeah, the blurring thing is tricky to do efficiently. That was why I was musing about the old transparency with Towards/Away trick. It would be nice if the renderer could possibly allow a screen space blur, but with proper AA samples.

      Maybe the jittering could be done in screen space as well, with the amount based on the circle of confusion (for DOF).


      Comment


      • #4
        Another thing I could have used recently...

        Ability to animate scale of the particles BOTH based on particle age, AND based on the master timeline.

        So, for instance, I could have particles animate up in scale over their lifetime, but I could also animate all particles to scale from their current size to 10% between frames 50 and 75, while still scaling up over their lifetimes.

        Independent XYZ offsets would be nice too. Like Vertical Offset, but one for each dimension, animatable of course. For bonus points these could be animated both over life, and over global time (like scale above). Global is most important on this one, though.

        Comment


        • #5
          Ah yes, these will be good to have, logging them
          Svetlin Nikolov, Ex Phoenix team lead

          Comment

          Working...
          X