PDA

View Full Version : Help with conveyor belt PFX sim



Chrusion
10-29-2012, 04:54 PM
I have a flat conveyor belt (with proxy collision) on which I drop particles on to (via gravity FX) from a very short distance above the belt. The particles have Z=600mm/sec motion upon birth and collide with the flat proxy collision object and slide off the rounded end of the flat conveyor belt on to another concave conveyor belt (with proxy collision obj).

The problem is I can't find a combination of parameters to get the particles falling on to the second, concave belt to NOT slide down the concave slope, cross the center line of the belt, and fly off the opposite edge. If I lower Bounce on the concave belt below 5% or so the particles don't collide. No matter what the value is below 100%, the particles slide down the belt at the same rate. I adjusted Fix and Friction up and down from 0 to 100 in various combinations and true to form, the particles did stick where they landed, but stopped moving down Z.

I simply want the particles to hit the second collision belt and not bounce or slide AND continue with their initial Z=600mm/sec motion (sliding along in a straight line on the second, concave belt). I use texture motion on the belt at the same Z=600mm/sec speed to simulate the movement of the belt carrying the particles.

Any tips?

cagey5
10-29-2012, 05:04 PM
I guess you've also tried variations on particle weight and resistance?

Chrusion
10-29-2012, 05:50 PM
No, because I want no resistance or friction to slow the initial sliding motion.

I also just tried Stick mode with 20,000% Bind Power and they still slide down the slopes. The particles fall in a band, like a water fall, the ones on the outside edges hitting the steeper slopes of the concave, slide down each side, cross in the center, and slide up the other side, then back down... rinse/repeat forever forming a repeating figure 8 flow of particles on the concave belt... not what happens on a real conveyor.

SplineGod
10-29-2012, 07:22 PM
how are you animating the conveyor belt? If you havent yet, try hitting the calc button.

Chrusion
10-30-2012, 09:48 AM
The objects are stationary. I am clicking calc after each parameter change. See attached scene...

Need particles on lower, concave conveyor to travel in straight lines the same as the upper, flat conveyor.

cagey5
10-30-2012, 03:14 PM
I doubt that can ever work. The belt is concaved and the particles will flow into the centre. I can't think of a setting that would stop that.

Chrusion
10-30-2012, 05:26 PM
Sheesh! Yet ANOTHER thing LW cannot do. This is getting frustrating. Year after year I keep running into LW not being able to do these very basic dynamic simulations.

Then it hit me! If particles can't be made to stick where they land and go in a straight line due to the slope of the collision geometry, then FAKE the slope by FRESNELing it! HA! Take that, LW!

So I do just that... stair step the contour of the conveyor into 9 flat steps (44mm deep x 9mm tall), just like a fresnel lens. But guess what? Yep, PFX totally freaks out on the new collision geometry by letting 50% of the 10mm diameter falling particles pass right thru (even with PFX at a bone-grinding 1mm resolution). Of the 50% that do collide and very nicely travel in straight lines down the length of the steps, particles that have a slight angle of travel suddenly veer off course when they are around 15mm away from the 90 deg edge/corner of the step it is nearest to and get sucked thru out the bottom of the mesh, falling to their death. <pulls hair out>

Wee....

erikals
10-30-2012, 06:20 PM
did you attach the correct zip file?
they look the same...

SplineGod
10-31-2012, 12:38 AM
Try this. Its a simple fix but should work.

erikals
10-31-2012, 02:09 AM
yes, the invisible object could also be animated and deformed (or even use clothfx / softbodies)

if so, you might wanna turn up the friction value to avoid the wavy motion in the beginning.

SplineGod
10-31-2012, 04:02 AM
There should be enough in the scene to get the basic idea. I threw in a collision object set to plane at the same height as the lower conveyor belt. I guess it boils down to the precise amount of detail needed.:)

erikals
10-31-2012, 04:09 AM
yep, i was also thinking about using Bullet as an alternative, but not sure how fast it would be.
(haven't tested bullet yet...)

might be an idea, if he has LW11 that is...

Chrusion
11-01-2012, 08:32 AM
Bullet isn't a particle system, so how to generate a continuous stream of instanced gravel objects such that they fall and "stack" on top of each other is the question. PFX sadly lacks what Maya N-particles can do quite easily (fill a volume via self-collision that calculates really fast).

I have to abandon, yet again, PFX as a gravel flow simulation system and just fall back to the good ol' low-tech days of using highly subD'd meshes with animated textures to give a rudimentary visual effect of gravel moving down a conveyor belt. But that's not all it's doing. I have 5 "sub" conveyors above the main conveyor, each feeding a different gravel aggregate onto the main belt, thus creating a layered mix at the end. I will use PFX for the "water fall" of gravel off these sub conveyors where each particle ages a whole 9 frames.

This whole concept just seems SO ridiculously simple for any 3D apps to do nowadays compared to the billions of particles under CFD control needed to generate hollywood style water deluges and wave spray, etc.

erikals
11-01-2012, 05:56 PM
Bullet can do it, though you need enough objects, so it's quite simple to set up.
the big question is the calculation time.

you can pile up particles in pfx, but as you say, not that fast though...

PFX needs an update, but not sure what NT will choose to do, probably use Open Source,
but the Bullet Source doesn't have particles i think...

Blender particles might actually be a good alternative here

a cheat is to use duplicated animated polygon bands, and use instanced rocks to attach to the vertices.

Chrusion
11-04-2012, 12:38 AM
Like clone a low-poly gravel rock/stone/particle to a cloud of points generally shaped like the mount/layer of material on the conveyor, then apply bullet to this object, allowing gravity and self-collision to "compress" it all together, then apply a force (wind?) to move everything along on the conveyor?

Still begs the question of how do you continuously supply/feed the conveyor with material at one end as the other falls off?