[novice] slow motion water and pyro sims 2 approaches

[art3mis]

Hi

Was hoping I could get either confirmation and or some feedback in regards to my understanding of slow motion simulations in Houdini.

For FLIP fluids there are essentially 2 approaches

1) Reduce Time Scale on the AutoDop network

2) Create your sim at normal speed, getting exactly the look you want. Then output with substeps, bring back into Houdini and retime using a combination of TimeBlend and TimeWarp SOPs.

 

For the past month(evenings and weekends as this is still a part time passion) I have been trying to get a usable result using the first approach

The problem with this approach is that the dynamics of the simulation change with reduced Time Scale.

Can any Houdini experts suggest why this approach is still often recommended as being 'simpler'?

 

 

 

Edited November 7, 2016 by art3mis

[dimovfx]

You could try with changing FPS, like if you are shooting with high speed camera. But then if you have any scripts its better to use Time instead of Frame.

[rbowden]

So it seems you are asking alot of questions and getting caught up on this slow motion flip fluids. You are getting caught up on what people are saying is 'simpler' also. In the end, it doesn't matter what method you use as long as you get the result you are looking for. What is it about option 1 that is not giving you a usable result? Is the sim blowing up? Are the forces messed up? Or are you just saying this because it doesn't look the same as a timescale? If you are going to go this route, unfortunately tweaking your settings is what you are going to have to do. If you are getting stray particles flying out and want to delete them, you can always blast those particles away post sim before you mesh. I don't think I have gotten a 100% perfect sim before (meaning that I haven't had to tweak it post-sim).

Have you even tried option 2 yet or have you just been super caught up with trying option 1, you haven't even tried it? If you have, what is holding you up? So many questions that need answering.

[art3mis]

Thanks Ryan

Option 2 is my next approach

My problem with the first approach is tweaking the parameters to get the look I want at the desired speed.

Somewhat like playing whack a mole, I determine the correct timescale, then focus on increasing say the feedback, or decreasing particle separation, or other parameters, then everything else changes!

I've spent weeks now(evenings and weekends) so far testing using wedges.

I assume that one bottleneck with option 2 is identical however, changing the particle separation will change the dynamics (effect of Feedback, amount of splash etc) so there is no way to really do a "low resolution" FLIP simulation and keep the same dynamics.

 

[rbowden]

Welcome to the world of fluid simulation. It is true that lowering the particle separation is going to change the sim a bit but, you can roughly get what you are looking for with a medium quality simulation (ie 8-10 mil particles). Get the timescale at the speed you want, then mess with the feedback if you need to. If you are looking for something to float in the water, a value of 1 should cut it on the feedback scale...Are you trying to achieve a certain looking splash? If so, look into creating your own force or velocity field and plug it in. You will be stuck in tweaking hell if you are trying to do something specific and doing a vanilla flip setup. I don't know specifically what you are trying to do but, here are some resources that could possible help you on the custom forces:

https://vimeo.com/129876655

http://forums.odforce.net/topic/15205-flip-fluids-and-curve-field/

https://www.sidefx.com/forum/topic/35468/

 

Remember that a FLIP sim is just particles and you can use pop nodes with it to shape it even more.

 

 

[Yon]

Looks like your interested in flip but  topic also says pyro. So I'll throw in the up res solver for retiming pyro works really well. I never adjust timescale for pyro, especially lower as it gives less time for the plume to evolve and looks bad. 

[art3mis]

What would REALLY help speed things up is if there were some way to use the SLICE tool to take a cross section of your FLIP sim at the final resolution-particle separation.

see my post here. Unless I'm misreading the documentation, it should be possible

 

[rbowden]

If you want just a small section of your FLIP sim for previewing, you can just change your domain to be in the area of interest. I tend to make my own domains anyway for this very reason.

[art3mis]

Thanks Ryan. Could you clarify? When you say ' change your domain to be in the area of interest ' do you mean simply change your camera's cropping, focal length - FOV?

[rbowden]

Uh, no.

I am talking about the domain of your actual flip simulation, your bounds. I can't upload anything on threads it seems like so, I uploaded a file to my google drive for you:

https://drive.google.com/open?id=0B6YJxsxjGJtFelI5Nl9sVW02aXM

In the flattank_inital node---wavetank sop, you will see that you have a size and center. When you use the shelf, those values are referenced in your flip solver under volume motion--- volume limits. What I do is make my own box and link that to my wavetank sop in this case. The new box will serve as my domain and I can size it and move it easily to wherever I want. The box I renamed to DOMAIN and is right beside that wavetank sop. 

Edited November 8, 2016 by rbowden
Forgot a sentence.

[art3mis]

Thanks! This should speed things up considerably.