Hello,

I have a fast moving smoke source, like a comet. However the camera is looking at the front of the “comet”, so the smoke trail is not visible, just the front face. I am having the hardest time generating smoke detail and visible smoke collisions in front of the comet. I am using the sparse pyro solver.

For collisions I tried sourcing in collision/collisionvel volumes and tweaking the collisionvel scale, I tried using the shelf deforming object collider, I tried high substeps on the sim (up to 25), still nothing.

For detail I tried using a gas vortex confinement microsolver with high scale(200), tried turbulance, disturbance, shredding, no help, all i see is a thin outline around the comet with hardly any detail.

I cant post images, but i have attached alembic of a sphere that follows a similar path, to get a sense of the speed, along with a camera.

Any suggestions or tips to generate detail, and get nice smoke collisions in front of fast moving sources, is much appreciated. I am a newbie to houdini (just a couple of years in), and look forward to learn what houdini offers to handle this.

You should check out some of Steven Knipping excellent tutorials on smoke simulations, I think it will shed some light on what you're trying to do. Totally worth the price of admission:

http://www.appliedhoudini.com/blog/2018/2/15/volumes-ii-simulation [www.appliedhoudini.com]

Navier-Stokes equations (pyro simulation) aren't really applicable to simulating comet tail, which happens in a vacuum and is not caused by collision.

https://en.wikipedia.org/wiki/Comet_tail [en.wikipedia.org]

Although maybe you mean something more like a meteor ablating in the atmosphere. Which again would be where standard pyro simulations break down as flow is super sonic and behaves completely differently than subsonic turbulent flow.

But maybe none of that matters in this case. Simulating fast moving colliders will always be a problem, as the simulations assume nothing moves more than a voxel per time step which is far too many steps for fast moving colliders.

Have you considered moving the refence frame to the sphere and camera, and moving the media instead? I normally achieve this with wind tunnel velocity, but I'm not sure if that works with sparse solvers.

It might also be worth considering a non-simulation approach to get interesting shapes on the leading edge, with the smoke simulation taking over for the actual trailing pyro smoke.

A little late, but I've had similar problem and this solution worked for me and I want to share it if someone look in future for it. The easy way is to feed your simulation geometry into cache and than import it in POP network(you can set pop source as first context geometry). Therefore you will have particle emission from your RBD particles or your big smoke object like meteor, plane engine whatever. Particles from Particles. The second particles that are from POP are making tails with their points ant this points are connected into your pyrosource node which is than connected into volumerasterizeattributes. In this way you will have prepared input for your volumesource node in your pyro/smoke dop netowrk with suitable shape . Don't increase substeps. It is too compute intensive.






with above mentioned method (one substep smokesolver)



and without:
particles from particles