How to approach this effect in Houdini

[~nature~]

Hi, everyone,

I found this Whirlpool on vimeo, which is generated and rendered by POV-Ray,

any ideas on how to approach this effect in houdini ? how to simulate whirlpool for a long time like this one without crashing the mesh itself.

[old school]

This technique is an old one I believe. Notice how the funnel geometry doesn't change at all as the textures are animated over uv space.

I wonder what was in the scene description file. A funnel shaped vortex geometry description perhaps?

I strongly suspect that the funnel geometry was encapsulated in the supplied scene description file.

Build the funnel in Houdini and make sure it has uv's properly formed then render out an ifd scene description file and then apply your own animated displacement map and surface map.

If you don't want the funnel geometry in the ifd, use a Delayed Load Procedural shader to fetch the funnel geometry at render time.

Edited April 18, 2011 by old school

[hopbin9]

yea, it's not a sim. It's just an animated geometry.

If it was a sim, it would look better then that

[Div]

Have a look at Miguel Perez' maelstrom thread here, hip file include

[~nature~]

~~~~~old school~~~~~

Thank you, old school, did you mean make uv form the vortex shape on the funnel geometry and flow or offset the displacement map and surface map based on such uv? I will try this technique,

do you have any ideas on how to simulate whirlpool physically accurately ?

~~~~~hopbin9~~~~~

Thanks,hopbin, I have tried the animated geo before, but to make whirlpool persist for a long time, I have to delete the bottom of the whirlpool geo or just cycle it. but if I animate the geo like this , the shading is a huge problem I cannot make the texture stick to the geo. Do you have any idea about it or how to simulate via fluid dynamics.

~~~~~Div~~~~~

Thank you, Div, I have looked into Miguel's version before, but again his Maelstrom cannot persist for a long time, after several seconds it will crash itself or distort a lot. So it is not a perfect solution.

I am really curious about how to simulate maelstrom/whirlpool physically accurately in houdini. but so far the texture seems the only solution to simulate non-crashed whirlpool which I will try. any further ideas would be welcome

Best regards,

Edited April 19, 2011 by ~nature~

[Macha]

I don't have a solution for this but here is an idea.

I used a vortex force with a smoke sim and then used its velocity to advect a dop particle system.

The result is stable but a bit boring and a little hard to preview (only points).

wirrly.hip

[pclaes]

How about a grid inside a sopsolver with a magnet sop rotating and pushing the points down. This would break the mesh after a while though. You can measure area of primitives and subdive if necessary to add more detail adaptively. Layer noises for small detail.

[~nature~]

I don't have a solution for this but here is an idea.

I used a vortex force with a smoke sim and then used its velocity to advect a dop particle system.

The result is stable but a bit boring and a little hard to preview (only points).

Hey Macha, thanks for your wonderful example file , and I think this method could be transferred to FLIP method, right? I am not so familiar with FLIP yet, I will look into it today and yesterday I found a whirlpool simulation done in realflow

it is based on realflow 5 grid fluid method which I guess is also doable in houdini using FLIP. Edited April 20, 2011 by ~nature~

[~nature~]

How about a grid inside a sopsolver with a magnet sop rotating and pushing the points down. This would break the mesh after a while though. You can measure area of primitives and subdive if necessary to add more detail adaptively. Layer noises for small detail.

Peter, thanks for your idea, but why bother dops with this effect? did you mean define the movement of the point grid in the sop solver and at the same time enforce divergence free in dops so that the points perform like water?

[~nature~]

Here is the procedural whirlpool pattern via the curl noise in houdini, but any ideas about how to make this pattern swirl inwards ? and the tunnel shape can be formed using magnet afterwards.

whirlpool_pattern.hip

[Macha]

Thats not bad already!

How about a trig expression to make it move.

I gave flips a whirl as well but they don't respond well to external forces. No matter how strong you move them every force seems to dampen out almost immediately.

whirlpool_pattern.hip

[pclaes]

Peter, thanks for your idea, but why bother dops with this effect? did you mean define the movement of the point grid in the sop solver and at the same time enforce divergence free in dops so that the points perform like water?

The only reason for dops is so you get the feedback loop going in the sopsolver. You could do this procedural, but how about a splash...? Like dropping an object into the side of the whirlpool: Think splatting some crater-like displacement pattern in the red channel, then the mesh gets "advected" by the magnet sop and the noise fields, so the red splat sticks to the mesh and whirls along. As a post process or as part of the main sopsolve loop, you can displace the mesh by that splat. If you do it as part of the main loop, that displacement will stick. If instead you reduce the red color over time and do a post process, you could blend in and out the crater displacement whilst it sticks to the mesh.

This is a complete hack and does not do an actual particle sim, you are advecting a grid, because all you really see is the surface. So for this hack you are not dealing with divergence free fields or SPH or FLIP, it is just a grid of polygons.

[~nature~]

Thanks Macha, trig expression helps a bit, but the inward movement feeling is not so strong, I have no idea about it.

Thanks peter, it's inspiring, I will try this idea.

Best regards,

[gany_1985]

Here is the procedural whirlpool pattern via the curl noise in houdini, but any ideas about how to make this pattern swirl inwards ? and the tunnel shape can be formed using magnet afterwards.

hi there ,ur method seems good..

can i use the same method on volumes so as to get a hurricane cloud.

if so kindly let me kno

cheers

ganesh