View Full Version : Fluid Swirl

08 August 2009, 06:06 PM
In my quest to better understand and control fluids, I did some tests on swirl. Swirl along with Temp turbulence is elusive. There's only 1 setting. More swirl, a ton more? what does it mean? How does it swirl? What is the result artistically and how far can it be pushed?

Those are artistic questions in a sense. The observations I've had is that it seems very depended on resolution for the "frequency" of the swirl, and how the magnitude and frequency/rez work together. I've attached a test image with a bunch of 2d fluids. The general working rez is 100x and high rez is 200x really generally to get good looking sims, for the most part. It's interesting how the swirl reacts differently when you go up to 500x and how small the frequency seems to be. Anyway, sheck it out, see what you guys gather from it.

08 August 2009, 06:29 PM
I don't have anything to add to this except the belief that Swirl is actually vorticity confinement talked about in this ( paper.

I think it basically finds the cells that are being damped unnaturally by the simulation then adds "swirl" to them, so when you simulate at a higher resolution there are of course more locations where it can add swirl. Though your 500x500 simulation would seem to show at high enough resolutions the unnatural dampening is as prevalent so the swirl isn't so turbulent. I could of course be talking out of my arse. :hmm:

08 August 2009, 06:53 AM

Thanks to Dave, it looks like more hi-rez is more beautiful. I remember you mentioned in your free tutorial about "Not more hi-rez that doesn't mean it'll be more beautiful". Does it react against each other?

In some ways, we have to recognize the truth hi-rez is better looking. But it builds on you make fluid fx in the right way.

Thanks to Dave again, I learn it.

08 August 2009, 11:46 PM

One other interesting approach to generating swirl I am finding is this. Which is better in generating swirl, to use bouyency to drive the fluid or a field? For example I have a scene with a skyrocket flying around a room and bouncing off the walls. If I attach a volume field to the rocket to create the horizontal force and add bouyyency for lift, Im not really getting the nice swirl. Whereas if I tilt the container on its side and use the bouyency as the horizontal force with some field driving the lift, then I seemn to get some nice swirl. But this method is inherently bugged since obviously the bouyency is acting in one direction only.

Be interesting too see some tests with this as well but I would assume that using bouyency would always be better as it kinda is localised each voxel containing fluid whereas the field is autonomous and doesnt need fluid.

08 August 2009, 04:39 PM
My one gripe with using swirls and turbulence is when doing shots with fire; they tend to clump into one mass.
Here is a quick sketch of this problem
A is a great simulation if one wants to sell a big fire, whereas B is a small scale fire.
However, maya fluids all tend to look like B and it happens when you slowly increase the turbulence and swirl value. This is a problem because swirls are supposed to add detail and not bring stuff together...

Honestly doing a single emitter as a test isn't going to cut it... to properly understand how swirl and temp turb work together one would need to have multiple emitters in different areas of the fluid box and see how the negative spaces get filled up...

08 August 2009, 09:54 AM

Great fx doesn't have one layer only.

08 August 2009, 07:27 PM
mine do ;) kidding. I understand swirl and turbulence fairly well, but I still study them for more insight. I can make a big fire (with one emitter oooooh) and use no swirl or some swirl and get to both good looking fires like "A". I'd say I can get fire to look 70-80% there (shot and subject dependent*). Double Negative's fire is probably 80-90% there, ILM 80-90%. How far can one get maya's fluid to look like fire? I'd say 85% to go from 80% to 85% is very hard. Each little scrap little trick gets you slightly closer.
In order to get those extra scraps of detail and control out of maya, we have to understand how not just swirl but all settings react under different resolutions. Really cool big fire needs quite high resolutions, probably the most resolution of any effect with fluids.

My new theory is to learn what settings are on a voxel scale and what are more global. Bouancy is global. Rase the rez and you are still in the ballpark of where you were. Temp turb is voxel scale. very different effects based on what rez the container is. This is all building an artistic understanding of fluids. meaning just get an intuative understanding of what settings should be where to get what you want. It's impossable to know how each thing will work in every case unless you are a scientist who understands fluid dynamics on a mathmatical scale... I am not. Anyway, I made a few more tests. I'll try to get them up for you all to see... or remake them.

*a good compositor can take 70-80 to 100 so it means you get it as good as you can and they take it the rest of the way. Hell this is all totally subjective to what I think those %'s are anyway so take it as you will.

CGTalk Moderation
08 August 2009, 07:27 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.