View Full Version : LW 10.1 Particles and collision objects

08-25-2011, 05:28 PM
Here's the scenario in LW 10.1...

I have a 6 vane rotary valve oriented horizontally as one collision object that's about 1' long and 1' in dia. (bounce; 100%; Object-Advanced; 1mm level; friction 50; roughness 500%; probability 200%).

A housing surrounds it with square openings at top and bottom as a second collider (normals facing inward; bounce; 150%; object-advanced; 1mm level; 0 friction and roughness; prob. 200%).

Next, I have a 100mm cube emitter popping out particles at 200/frame with -200mm X motion, particle size = 5mm dia. +/- 2mm, self-interaction (Push mode; force 200%; Viscosity 10K%) enabled and the other Interaction = None. All other params at default.

Last an FX Gravity at -2m Y. ALL FX using default group.

The vane rotates in steps... 3 second pause with a set of vanes pointed up to catch the falling particles like a "V" followed by a 1 second rotation to bring the next set of vanes into alignment.

My goal? To get a "V" to fill up with dusty dirt, then have it poor out when the vane rotates past level.

My problem? 20% or so of my particles randomly shoot thru the "walls" of the colliders during Calculate... even if they are just sitting on a horizontally oriented vane like this: -----0-----

The remaining 80% or so jiggle around while particles collecting in the adjacent "V" trough shoot thru their vanes as well.

I've tried FOUR different geometries for the vane object. 1. very simple polys with sharp-edges. 2. subdividing it into a julienne mesh. 3. applying Rounder to fillet all edges of the simple version. 4. julienne'ing the rounded version it into a mesh. ALL produced the same results... instantaneous and constant contact collisions NOT 100% respected!

So, what is the secret to prevent particles from immediately ignoring or eventually ignoring collision objects? Yes, I've played around with the FX Browser resolution option... 100mm default, 10mm, 1mm. The higher the rez (smaller #), the worse it got (and of course slowed the calculation way down).

And what is the secret to get self-interacting particles to stack vertically inside a collision object like pellets 6" deep?

Anything over 2 particles high and they all just ooze together in a quivering mass of jello, intersecting and overlapping, sticking to the side walls of the "V" and then bursting forth into an oozing pancake when the V rotates so a vane is flat/horizontal. My client does not want jello dirt. They want dirt that tumbles, not oozes.


08-25-2011, 06:00 PM
Loop this quicktime to see what I'm saying...

Mr Rid
08-25-2011, 08:51 PM
Archaic LW particles may not be up to your task. Turning the resolution option down should prevent particles from passing thru geometry.

Cranking up Force should make particle self interaction less slushy.

The jittery thing is unavoidable.

Calculation time may be prohibitive for a high enough particle count you may need to make this convincing. Self-colliding PFX' bog down quickly.

See if this helps at all-97779

There is the Dstorm Liquid pack thing that I tried once, but I dont know if is working in LW10, or if will solve your issues.

08-25-2011, 11:47 PM
Liquid pack is good, but only works in 9.6, cost about $200 or so if memory serves.
had some leek issues too, but a bit better than lw.

08-26-2011, 12:07 AM
Thanks for your test scene Rid!

Seems that my geometry must be messed up in some invisible way. Your vane valve doesn't "leak" particles like mine does, so I rebuilt my vane and substituted it for yours and the collision is 100% now. I wonder if the "leaking" was due to my collider having side walls, whereas your scene uses FX Collision planes, instead.

Anyway, I tripled the number of particles and see that they are sort of stacking up higher with your 1000% force and viscosity for push self-interaction, but, as you said, they just turn into a quivering, jittering mass... especially the top layers. The bottom layer touching the collider (vane) stays relatively stationary and fixed in place (somewhat), while all the others above them jump all over the place.

Interesting to see too that the bottom layer is most composed of the smallest particles, just like shaking a box of cereal... all the large flakes rise to the top and the dust to the bottom... hehe.

So, I guess there's no workaround to prevent this (quivering/giggling). Client isn't going to be happy at all.

One thing is for sure... LW PFX is a thousand lightyears behind Maya n-Particles, which is probably 100 lightyears behind Lagoa Multiphysics (love to get my hands on that.... sheesh).


Mr Rid
08-26-2011, 03:54 AM
I dont know if Blender fluid is a viable option.

08-26-2011, 07:40 AM
I'm just going to have to "fake it" with geometry and morphs for filling/emptying the vane cups and use pfx for the falling dust/dirt.

I should have spent my time working on doing it this way rather than wasting it on all this pfx nonsense.

Now I know why I get evil glares from most of my fellow Maya artists when ever I say I use only Lightwave... that or snickering chuckles that turns into full out bellowing laughter halfway down the hallway after they exit the room.

I guess PFX was designed ONLY for explosions, water fountains, smoke trails and the like, leaving we volumetric, fine-grained, real physics simulation special effects artists in a dark empty void. [ Newtek < ping? Hello? ]


08-26-2011, 10:47 AM
...Blender should be able to do it, LoL..! :]

looks nice actually... :]

hmm,... maybe it could work... ?!
export the particles from Blender to Lightwave via MDD...

Mr Rid
08-26-2011, 05:13 PM
I know a LW feature you can rub in their Maya noses... Zoom Factor! And you can select up to TWO specific lights to shine on sprites! Undo? pfff, thats for Maya weenies. Why would anyone NOT want to have to flip back and forth between two apps to modify a model?

08-26-2011, 05:24 PM
I know a LW feature you can rub in their Maya noses... Zoom Factor! And you can select up to TWO specific lights to shine on sprites!

We had a 6 light capable, almost realtime tool in Maya that was used for this project (http://www.youtube.com/watch?v=WOK7r1pa74M). All smoke effects done with it. :)

What is it about Zoom Factor that is so great compared to equalent channels in Maya?

What I would suggest when it comes to "rubbing in the face" regarding LW-features not related to rendering, is the FPS-esque camera and how easy it is to use only your mouse or tablet to create "handheld" cameras or the endomorph system in regards to changes of the mesh.


08-27-2011, 02:35 AM
animatable UV in Lightwave makes dust of the same "feature" in Maya... ꞉P

08-27-2011, 06:47 AM
Very nice zoom in of the orion Nebula 1 minute in at the clip..cool:thumbsup:


08-27-2011, 07:00 AM
Very nice zoom in of the orion Nebula 1 minute in at the clip..cool:thumbsup:l
Thanks. Did you notice the use of that age-old, archaic, "volumetric" rendering technique? Can Maya do the same? hehe

Luxigons are cool too. Does Maya have those as well? :-P


08-27-2011, 07:33 AM
Thanks. Did you notice the use of that age-old, archaic, "volumetric" rendering technique? Can Maya do the same? hehe

Luxigons are cool too. Does Maya have those as well? :-P


I donīt know really how you approached it, can only speculate in projected images on geometry, or mapped?

But it got a nice depth to it..by the soft look of it I could almost think that you used a point cloud sculpted cluster and mapped the real orion image map on the Hv sprites?


08-27-2011, 08:00 AM
I have SO fooled you! WIN!

It's all geometry. The neb is a dense mesh of tris, frozen from sculpted subpatch quads with higher density in the center. Shaped based on the mesh used by the scientist who made the true volumetric 3D orion neb several years ago using supercomputers.

I then displaced the mesh with simple turbulence using the normal displacement plugin and Saved Transformed. Used this to create an endomorph. Used morph mixer to ramp from 0% to 100% in one frame with repeat post behavior, BUT I edited the envelope to have a very slow ease in and a hard, almost logarithmic ending keyframe.

This is the classic "spinning light trick" we used to use to get soft ray-traced shadows from distant lights eons ago. The key is motion blur to "march" the surface thru a texture in one frame to create a psuedo-volumetric effect.

So, in this case, photoreal motion blur with its stocastic noise distribution of pixels and adaptive sampling AA to smooth them together, combined with the progressive displacement of the surface created the fluffy cloud effect and the image mapped mesh. And because the displacement advances "faster" towards the last inter-frame step, the composite image has less "density" at the furthest extent of the displacement and more at the start where the displacement happens very slowly.


Mr Rid
08-27-2011, 11:41 PM

What I would suggest when it comes to "rubbing in the face" regarding LW-features not related to rendering, ...


I know. Job listings in the last two months:

CGsoc- 8
LW- 0

Maya- 16
LW- 0

Creative Heads-
Maya- 100
LW- 4


08-28-2011, 02:35 AM
seriously David, add and learn another app,...

Mr Rid
08-28-2011, 03:30 AM
But I do, and I have.