Adaptive Grid Expansion Based on Velocity Channel or on Particles

[Joelaff]

Would be cool if the adaptive grid expansion could be based on the velocity channel of the grid. This is the default for the cigarette smoke setup (to only save velocity). Or even better it could also be based on particles as well. So anywhere there are (e.g. drag) particles the adaptive grid would expand.

[Svetlin.Nikolov]

Oh, it's already possible to use the Speed channel as a threshold - I actually use this often and recommend it in case of complex fire/smoke simulations that have only fire in some parts of the grid and only smoke in others - setting the adaptive grid channel to Speed and using a low threshold, so movement in the grid would cause it to expand.

However, taking a not about the drag particles expanding the grid - this is the first time we get this requested, but really makes sense. I was also looking into allowing drag particles to collide with geometries when I was experimenting with creating ballistic particles last year. Hope I will have time to come back to this soon.

Cheers!

[Joelaff]

I saw speed in there. I didn’t try it, though.

Would I have to the also store speed in the grid files written to disk? Or will that work without storing the channel? I may have made a stupid assumption that it had to be a channel saved to disk.

Does the cigarette smoke setup NEED to store velocity, or any grid channel on disk? The particles have velocity turned on. I wasn’t sure if I needed the grid velocity for something, so I left it on, since it was on in the preset that I used as the basis but my sim. (As usual, given very little time to complete a shot, lacking little time for testing…)

I don’t normally use the presets, but I thought I would try them, and that one got me close right away.

Adaptive grid based on particle would still be cool, though I suppose speed would work in most instances. Thanks.

[georgi.zhekov]

No need to store the Speed chanel in order to use it for the Adaptive grid. It should work out of the box.

And in cases where you export the Velocity grid channel - the Speed channel can be derived from that so you can get away without exporting the Speed channel.

For the cigarette smoke - if you will be using the preset as is and the drag particles have the velocity enabled in the Source settings - you probably don't need the grid velocity. The particles should render with motion blur even without the grid velocity channel.

[Joelaff]

Thanks for the info, Georgi. That's exactly what I wanted to know.

[Svetlin.Nikolov]

Indeed, just wanted to rephrase this so it makes more sense in general - nothing you choose in the Output rollout should ever affect the behavior of the fluid solver in any way. The data saved to the caches would affect the preview, the rendering, as well as other interacting simulators, but not the simulator's own simulation - if a channel is not checked in the Output rollout but the solver needs it, it will take care to enable it under the hood

I am paying a special attention to this topic since there are misconceptions about it, and there were such even in the team, so we did a pass through the entire online docs and corrected several places that claimed the Output rollout affected certain simulation options, also verifying that this is not the case at least with the lasted Phoenix versions.

Cheers!

[Joelaff]

Thanks for the clarification. I thought that I didn't need to save grid channels, but since the cigarette smoke preset defaulted to including the grid velocity I wasn't so sure. (And I was in a hurry due to a deadline, otherwise I would have tested.)

I thought the particle velocity was independent (as it is), but I also understand that you can also do some cool things with grid channels at the shading stage where particles are concerned.