What is legal in a SOP solver's SOP chain? (Could be an RFE)

   2275   0   0
User Avatar
Member
14 posts
Joined: Dec. 2007
Offline
I am trying to use the Pops solver in Dops, basically to get levelset collisions.

It works. But what I am stuck on now is trying to get data that I can see in the details view (say, ‘Impulse’), stuck back on to the appropriate particles.

There don't seem to be operators or expressions to apply dopfield-type data back onto geometry. It seems that for this kind of thing, you would try a multi solver/SOP solver setup.

So I can do something simple, like change my particles' colors with a point SOP. The basic framework seems to work. But when I try to do something useful, I seem to just make Houdini crash. One problem with this is that this transient collision data is sorted only by some point number scheme, so you need to do some kind of id lookup to get it to go to the right place. So the most direct thing to me seemed like writing a python SOP to maybe build some useful data structure for the relevant dopfield() fields and then applying the attributes to the correct particles. But my first simple python sop (which tested fine in a plain old /obj/geo/ application) just crashed houdini in a sop solver.

I also tried Chops, and using the dynamics chop seemed promising, but trying to use a geometry chop in this (sop solver) context also crashes Houdini. The closest I got here was using data that I dug out with the dynamics chop to generate a new particle system, and then attribute transferring the relevant data back onto the original system. This actually kind of worked (well, sort of).

What is needed (i think) is a sop solver-friendly sop (or pop solver-friendly pop?) which can be pointed at a bit of dop data and will place it on a point as some user named attribute. And/or some kind of HOM support for doing this kind of thing in a script solver (I poked around there too - couldn't find a way to alter ‘Geometry’ type data).

Any suggestions? I guess I might try VEX next, but there seems to be an underlying ‘touchiness’ to the Sop solver, so I am not particlularly confident…

This was with various recent cuts of 9.0/9.1 on Linux.
  • Quick Links