moving objects with particles?

[wannabeArtist]

Hello,

I'm starting to learn some basic fluid simulation with the great SPH addon - in a more general sense, is it possible to use particles as a force to move objects?

I don't know much anything about this yet, but I would guess it's not impossible to pick up the velocity vectors and mass data from the particles and use it to "push" around objects - think of boxes in a river, for instance. Is this doable?

[wannabeArtist]

I thought I should elaborate this a little...

Here's a little test scene I made, it has a sphere in a box and a disc acting as the fluid (particle) emitter.



I made the sphere an active ridid body and the cube a passive one and added gravity using the regular simulation tools (is this doable with ICE?).
So, the ball falls nicely into the bottom of the box.

Now, what I would really like to see, is the fluid (the particles) to lift the ball from the box.
So far no cigar - the ball stays in the bottom, regardless of it's tiny mass, so something is missing here - any thoughts?

[Tekano]

hi - the only way I see this interaction happening the way you want it to keep the sphere active RBD and turn on all the particles to be RBD too - this will of course affect the SPH sim - so you would need to turn off self collision for the particles RBD and only collide with the sphere. again there is no guarantees this would work properly.

[wannabeArtist]

Thanks,

I tried making the sph cloud rigid body earlier today, but I think I messed something up, since it made XSI crash - I need to investigate a bit more on that.

I also posted this same question to the SPH group, in case somebody threre has tried it before

[owei]

Hi..!

Take a look at this thread speacially at the swimming yellow ton:
viewtopic.php?f=19&t=17 .

Just use this compound I called "swimmer" to do some of the work you need. Averageing the velocity of the influencing particles and puting this on the object would be no big problem to implement. I´m very busy at the moment, so I can´t do that at the moment. Remember, that you have to switch on the unsupported kinematics feature ,)

cheers,
oliver

[wannabeArtist]

Hi!

Great work! - I'll definitely will check that out!

However, in this project my real goal is not the ball swimming in the box - it was more like a test, to see if I can make the partilces move an object.
Rather I'm trying to model a simple hydraulic device. I want the particles to "push" open a little latch inside it and if I have understood correctly so far, I need the rigid body to do this.

That Swimmer thing is fantastic, though - I need to come up with another idea to try that out

[owei]

In my eyes you don´t need RBDs to do this...yes, of course you can push things using particles an a simple ICE-Tree. That´s what this compound does...it also rotates it...I think, this compound can be modified the way you like very easily..

cheers,
oliver

[wannabeArtist]

Hello again,

Well, that shows what I know - I'm really just beginning with ICE

I'm trying to learn to use your compound with the ball test right now, but there's one thing I didn't quite understand yet - it's the surface input: I tried to mimic the example in the other thread, but simply getting the pointpositions from the point cloud wont do the trick - perhaps I need to derive the surface somehow out of the point cloud first?

Also, what's the unsupported kinematics feature, how do you turn it on?

Thanks again!

[owei]

Hi..!

Add following line to the "sentenv.bat" found in the application folder of Softimage:
set XSI_UNSUPPORTED_ICE_KINEMATICS=1

There is no need to pipe in the pointpositions. That does the compound for you. Just pull in your pointcloud into the ICE tree and connect it via the violet(geometry) connector a shown in the example.

hope this helps ,)

oliver

[wannabeArtist]

Thanks, sorry to bother you constantly with this

I figured out the kinematics bit, and found a nice video describing it: http://www.vimeo.com/1658393

Still not out of the woods with the surface, though.
I tried plugging in the point cloud directly, but the swimmer stays red with an error saying that "required attribute does not exist".
Now, this is an SPH point cloud - does it make any difference or what am I missing here?

Oh, and inside the Swimmer, it's the pointnormal node that's red

[owei]

...well, the compound expects an geometry a input with normal vectors to compute the rotation. Just go into the compound and disconnect the "set data node" from the branch that turned red. It should work after this manipulation...

[wannabeArtist]

Thanks again,

I disconnected the three lowermost nodes from the rest, the get pointnormal > get array average > direction to rotation - branch.

Swimmer turned blue now, but still no cigar- ball wont move.
I'm suspecting the problem is now my positioner input. I plugged in the same sphere as for the swimmer, but I see the example has "boje" for positioner and "boje.boje" for swimmer - so what's boje.boje

Also, I guessed that the swimmer needs to be last on the execution list, after the fluid simulation - did I got that right?

[owei]

...just append your scene..I´ll have a look at it..can´t tell what´s going wrong then... ;)

[wannabeArtist]

Thanks a ton,

Here you can find the scene and for a quick look, here's the ICE tree:

[owei]

...first, you should DELETE the sphere and make a new one, because you had RBD propertys on it. That does nor work with ICE..
secondly, create a null and use ist as a postioner object. It gives the xz position of the swimmer-object...canßt build ist for, no time

[wannabeArtist]

Thanks for the tips!

Darn, I just don't seem to get this right:

Now I have a new sphere, with no RBD properties and the null.
I'm not sure if you meant the null to be connected with the sphere outside the swimmer? I tried with and without, but either way my ball isn't moving yet.