Dissipation effect not looking good

[stevesideas]

So, I've created a steam animation which comes from an iron at high velocity. That part is looking fine but when the smoke/steam dissapates I get strange blobs of varying opacity. It resembles smoke as oppose to steam. Is there a way to do this? My settings and the result are attached.

[Joelaff]

Adjusting the smoke curves (using the smoke channel rather than simple smoke) can help, as can lowering the minimum visible opacity for simple smoke. It seems to cut off too sharply in the thin areas, leaving the stringy / blobby areas you describe.

I, too, am interested in any other ides on this because I agree it gets kind of "stringy" sometimes during the dissipation, and doesn't always look natural. I have on many occasions restored to blurring it in post during this dissipation.

[georgi.zhekov]

Indeed Joelaff's suggestions for the curve adjusting and the minimum visible opacity are the way to go here.

[Svetlin.Nikolov]

Also, please note that the viewport GPU Preview might not look exactly like the render, so try rendering it out and compare just in case. If I remember correctly, for the viewport GPU preview the Minimum Visible Opacity is always forced to 0.001 internally in order to avoid performance suicide with large caches that might have a lot of empty space.

[stevesideas]

Thanks for the tips. I will try this now and let you know how things go. Thanks.

[stevesideas]

This is what I'm getting when adjusting the curves using the smoke channel. The GPU display is very different from the render but also, the opacity just doesn't look that natural. It does look quite blobby to me.

[Svetlin.Nikolov]

Try adding some spikes to the Smoke Opacity curve. In addition, with Phoenix 5.10 that we released today, you would also be able to use V-Ray IPR CPU in order to get a more precise preview than what the gpu preview could offer.

[stevesideas]

Here are the tests. The first test uses multipass and the dissipation is quite nice but you lose the movement from the initial fast velocity of the jet or steam. The second is forward transfer which is probably about the best result BUT you do lose a lot of detail in the dissapation.

It would be nice if you could blend between forward and multipass over the course of an animation.

The last animation is reduced dissipation and also using the method you suggested above. Now there are lines in the smoke...obviously where I've reduced the opacity for that particular temp. Not really sure how to get the result I want with phoenix.

What do you think, does it look like steam in the first two? Maybe the first one is fine actually.

https://stevesideas.net/client/chaos/iron_tests_.mp4

[Joelaff]

I assume this banding/ringing you see here is to due to adding some kind of spikes to the smoke curve?


That's obviously suboptimal.

Can you post maybe one frame of the .aur sequence with a simplified scene with just the smoke, camera, and lights (no hero model)? Maybe someone could take a look at that. (I will if I have the time, as I am sometimes fighting the same issue.. The issue I am often fighting is the kind of stringy look you see in the first shot here:

[stevesideas]

Yes, the banding is because of the spikes in the opacity curve.

The stringy you've posted is what I was getting also. Are you using forward transfer? This effect was reduced when using multipass but I lost the velocity of the initial jet of steam. I'm trying to use particles but need to work that out....hopefully Tyflow is the answer!

I'll post something shortly.

[Joelaff]

I just ran a test and was able to get pretty darned smooth dissipation with a Minimum Visible Opacity of 0

With default MVO of 0.001


With MVO of 0.0


Those are just IPR renders.

Slightly better still using the Spherical sampler.

Phase function is nice to get some direction to the scattering if need be:



Note that is just 720p. So blowup in the VFB. It also has filmic tonemapping ON, but it seems pretty smooth.

[stevesideas]

ok. Thanks. I'll try those settings. That does look a lot smoother!

[Joelaff]

You may not need Raytraced or the phase function. That was just me playing with it since it made the backlit steam/smoke look better in that test.

[Svetlin.Nikolov]

Wooo, sweet! Thanks Joelaff! I was also going to link some resources for the smoke curve since indeed it ultimately produces manfolds that are closed surfaces and it needs some experimentations to make them not that obvious, for example, check here: https://docs.chaos.com/display/PHX4M...ing+Thin+Smoke

Also, another idea I had was to combine Forward Transfer with a Drag force to slow down the smoke and keep it near the iron. I think that Forward Transfer really helps show the jets from each nozzle much more distinctively.

Cheers!

[Joelaff]

Sounds good. I didn’t mess with the simulation settings much. I think I used 3 steps per frame, though not sure that helped anything. Rest were set to default other than dissipation.

Lowering the Step size below the default 90% as your tutorial link suggests probably would help, though. I didn’t try that.

My grid wasn’t very dense either.