particles driven by vectorfield

[resonanz]

I am looking for a way to control particles in a way that they move around obstacles with different shapes. Because the obstacles vary and the particles should slide on a surface I do not know how I can achieve this. Searching the web I found some methods which use vector fields to drive particles. My question is now how can I create a field like in the links below but rather for more obstacles?

Any help would be appreciated!

link1

link2

obstacles.hipnc

[Macha]

Here is a simple starting point.

I think something a little more sophisticated could be done with a pointcloud setup.

Or better, just the old get-vel-field-from-dops method.

vecfield.hip

Edited January 20, 2011 by Macha

[resonanz]

Thank you Macha for the file! It helps me to understand how I can drive particles by vectors. But my problem still is how to create a field around more than one obstacle and tangent to the surface.

Do you believe this is possible with your second method?

Thank you!

[Macha]

To do this properly in sops might be very difficult because -I guess- you need a correct gradient flow of the object, some other flow of vectors around it. Moreover, it needs to be directed. Mere tangents will probably work somehow, but not really correctly. You could try edge force in the point sop and see how that goes.

The dops method definitely works but it is the other way around from what you describe. The field isn't there from the start but rather the smoke (in that case) creates a velocity field that you then bring back into pops and advect your particles with.

So the question really is whether you want a workable solution or wether you are just curious how to do it the way you described it. I think both are possible ultimately, if you are smart enough (so, not me then, yet)

Edited January 20, 2011 by Macha

[resonanz]

To do this properly in sops might be very difficult because -I guess- you need a correct gradient flow of the object, some other flow of vectors around it. Moreover, it needs to be directed. Mere tangents will probably work somehow, but not really correctly. You could try edge force in the point sop and see how that goes.

The dops method definitely works but it is the other way around from what you describe. The field isn't there from the start but rather the smoke (in that case) creates a velocity field that you then bring back into pops and advect your particles with.

So the question really is whether you want a workable solution or wether you are just curious how to do it the way you described it. I think both are possible ultimately, if you are smart enough (so, not me then, yet)

Thank you Macha for your answer!

I would like to do something I have seen in this paper link

It is on page 16.

[Macha]

A practical and artistic way would be if you try something like my above file, but use a comb sop to brush your vector field the way you want it.

[petz]

as macha said, the question is mainly what´s the intention of doing this!

if you are interested in the effect of flowing particles around objects you don´t need vectorfields. and if vectorfields are the thing you will get familiar with then your example is probably not the best. but reading through your posts i guess the question is more related to the second point, right?

if so, one possible solution would be to compute "pseudo" bent normals pointing in the average direction of unoccluded samples. should work well if your pointcount is not too high. another possibility is using an attribtransfer-sop what will be faster but less accurate.

hth.

petz

vectorfield.hipnc

[Macha]

Ah, as usual, a superlative reply . I didn't know we could do it that way. Very interesting, and useful!

[resonanz]

Thank you very much Petz, I was looking for exactly this sort of example, although I do not fully understand how you get the vectorfield