pop sdf collisions in dops

[pgrochola]

Hi,

I'm doing POP collisions inside of DOPS.

Does anyone know how if it's possible to have a different collision offsets for each particle?

As if the each particle was a different sized marble ball when it hits the dops SDF?

Thanks!

[Macha]

My Houdini-instinct says "try the point collider and colliderlabel dop" but I can't seem to get it right. But maybe it helps?

I have another idea but I don't know if that would be possible, but perhaps we could store the offset in a field and somehow use it that way. This would be a more sophisticated solver setup and I don't know how to do it (yet).

Edited March 14, 2011 by Macha

[~nature~]

Why not do the SDF collision in POPs and import them in DOPs, the per particle volume offset is much easier in POPs by using $PSCALE atribute.

the setup is using volume sample compared with $PSCALE to test whether the particle has collided or not, if collided, reflect the velocity.

As for the DOPs, my initial guess is using VOP force to get acceleration and finally change the velocity at the hitting point, but reference the $PSCALE from the pop object geometry using point expression. I do not know whether this would work, I will test it when I am with my houdini

Best,

[Macha]

Great idea Nature! You're a genius!

Why did I never think of that, sdf collision in sop pops! Beautiful! Below is a file with a -somewhat rubbish- per-particle offset.

I must admit, I do like pops in dops as well. It's fast, realistic, and we can scrub it.

That may solve many collision probs I always fought with.

You just made my day!

Edit: disregard the popinput, it's a leftover from a debug.

offset.hip

Edited March 15, 2011 by Macha

[Macha]

I did a video of it for anybody interested:

[Hazoc]

I did a video of it for anybody interested:

Thanks Macha. This might be useful with my melting experiments:)

There will be lots of particles sliding down object's surface and yesterday I was planning to try something like this SDF method to avoid real collisions.

Edited March 15, 2011 by Hazoc

[~nature~]

Hey, Macha, thanks for the example, I've also tested this idea in DOPs , but I cannot come up with the accurate solution like POPs using VOP force only(gain momentum to change velocity). POPs should be the final one.

Cheers,

[Pazuzu]

Thanks Macha great example!!!

It's possible that you share the file when you calculate the flow direction of a mesh with the help of a SDF?

Thank you.

Edited March 15, 2011 by Pazuzu

[pgrochola]

Wow, thanks everyone!

very useful info.

Thanks Macha great example!!!

It's possible that you share the file when you calculate the flow direction of a mesh with the help of a SDF?

Thank you.

[symek]

We use this method a lot for some time now, very handy, but... shouldn't be there volume gradient from file VOP, not volume sample from three prims (there is only density in a sphere)... how could this possible work? (it does).

[Macha]

The volume density changes. It changes in all directions and we can measure this with 3 components (gradient). Since it changes along the normal (I think that's how sdf is calculated, distance from the surface?) it would represent the (unnormalized) normal directions. The volume analyze sop provides that gradient volume once (no need to recalculate every time). Previously I calculated my own curvature but then I saw Jeff use that node on sidefxforum and now I think it's one of the most useful new things in H11!

Bah, maybe I am wrong, I dunno.

Edited March 18, 2011 by Macha

[symek]

All is correct, my fault, I haven't notices Volume Analysis SOP at all. I have thought you're sampling single density fields three times, which obviously isn't the case! I usually use volumegradient()...

cheers!

skk.

Edited March 19, 2011 by SYmek