Smoky sims

Here’s some of my weird smoke tests compiled for your enjoyment :slight_smile:

Absolutely loved it! Gorgeous! Thank you for sharing!

can I ask what you tend to use for domain resolution / hi-res settings?

Thanks :stuck_out_tongue:

It’s often been around 80-120 domain res + 3-4 hi res. It really depends on the domain shape (the less cubic domain the less actual cells) and scene. For more accurate object collisions/emissions you tend to need to raise the base domain res but then you might not need that many hi res levels. I often just bake few frames, look at the ram usage & simulation time/frame, stop, and then test render one frame to make sure I didn’t produce some ridiculously big voxel data to render.

As you might have noticed, I actually didn’t use shaded rendering for most of these, but some combination of smoke emissions instead :slight_smile:

Truly enjoyable simulations to watch, Tame! Well done!

I was wondering, would you have any advice for achieving the “emission effect” of a tank firing? I’ve noticed the initial gasses expel from the barrel, and then often seem to hit an “invisible wall” (I’m assuming this is due to the huge difference in temperature outside the barrel), many times forming a disk shape. I thought using an obstacle to persuade the smoke to maintain this shape might work well, but do you have any other recommendations? I would like for it to look as natural as possible, without any overly-smooth areas. It’s for an extrememly slow-motion shot, so details will be very noticeable. Here are some references which help to show the effect:

Hmm, really interesting.

I did some quick tests and looks like something similar can be done with some tweaking.

It really comes down to which kind of shape and behavior of the “explosion” you want to achieve…

I can’t give much advice, but most likely you need to use multiple types of force fields (and keyframe various parameters) to shape the expansion to your liking. There’s also a force field object called “drag” which you could give a try. You could keyframe that strong in the beginning to drag down the velocity of your ever fast expanding smoke, and then soon lower the intensity to allow smoke drift forward etc. You also might need to use multiple smoke emitters, one for the large blow and other for some trailing smoke behind the projectile etc.

Anyway, I don’t think you can get very believable results with stopping the smoke with “a wall” :slight_smile:

Nice smoke tests. I like the helix rising one the best.

I am looking at your posted machine specs. So does the i7 quad perform well for these simulations?

Thank you for your suggestions, Tame! I will try them out. It is much appreciated, I can only hope that my results will be as cool as yours here are!

Well yes :stuck_out_tongue:

All of the newer i7 (Sandy Bridge and up, and maybe even i5) desktop CPUs should be good. But yeah, it seems that simulations also party rely on the single threaded performance… Only when you have more hires etc does the cpu spend more time at 100%, then having more cores would speed the baking process a bit more…

With my cpu I can most of the time press play and bake some real time preview, without hires up to ~120 domain at 1fps, depends much of the scene…