Deforming vs RBD Object in FLIP sim

[art3mis]

In many threads I've read here as well as in recent webinars it is recommended to use the Deforming Object shelf tool when setting up your DOP network in a FLIP sim. And then always turn off 'Use Deforming Geometry'

However in a simple FLIP sim I have with falling objects into a FLIP tank RBD Object is the only option that seems to work. If I use Deforming Object I can't get the gravity force to take effect.

Can anyone tell me in a nutshell what the advantage of Deforming over RBD Object shelf tool is, and how I might enable the gravity force when applying Deforming Object to any geometry?

[cudarsjanis]

Deforming Object shelf tool makes a static object and static solver in dops. RBD object makes RBD packed object and RBD solver. 2 different things.

If you want your object to interact with your flip -  like floating use RBD, but if you have animated object and animation is not going to affected by flip, but you want the flip to be affected by your object use static solver.

Anyway I would suggest to try to build your dop network yourself, then you get more understanding what and why is happening.

[art3mis]

Thanks. You're right of course, re building DOP network from scratch.

On my list of things to do!

Still wrestling with the finer points of creating FLIP sims at reduced scale, simulating surface tension,  and understanding the relationship of particle separation to things like geometry and FLIP tank size.

The documentation really only skims the surface! (pun intended)

[cudarsjanis]

From all the FLip tutorials I found this one to be the best. Yes it is a bit outdated, but loads of things still apply.

https://www.cgsociety.org/training/course/fluids-fx-using-houdini

 

For more basic stuff Digital Tutor ones for flip is the right place to start.

have fun

[art3mis]

Thanks but $699 ???

[cudarsjanis]

I know it`s a bit much. Just saying it was very usefull for me.

I purchased Pyro tutorial from Helloluxx guys. It was very good.

so I`m guessing flip one is good as well. that`s 39$

http://helloluxx.com/product/houdini-training-learn-jumpstart-vol-5-intro-to-flip-fluid-simulations/

[Martin47]

take a look a this Webinar from jeff Wagner, he talks about that here.

 

[art3mis]

Yes have the Adam Swaab tutorial. Great teacher.

Will have to rewatch the webinar. yes Jeff talks about separation, but doesn't go into enough detail IMHO. ditto his followup webinar.

I downloaded a .hip file from another thread here which has an amazing glass pouring sim but the author or artist only used a separation value of .001!

Of course when I tried such low values Houdini threw up and my machine crashed!

My fliptank and the scale I was working at was much too large!

One thing that still confuses me when it comes to scale is whether a 10 meter wide FLIP tank with a Particle Separation of say .02 should function or simulate the same way as a 1 meter FLIP tank with a Particle Separation of .002? Much more testing needed.

 

[rbowden]

Small scale and large scale fluids use different sets of values for a good reason. It is easier to control smaller scale fluid than  large scale fluid. To answer your question though, a 10 meter tank and a 1 meter tank will simulate completely different. Don't get caught up in the particle separation either because it is going to be different depending on the type of sim that you have and how deep it is. The biggest thing is to pay attention to the amount of flip particles you are dealing with.

Jeffs webinars are great for beginners and for people wanting to understand more about the parameters of the flip solver. What more detail are you looking for in tutorials? Unfortunately most of them out there are only going to go over the basics, go over specific parameters, or how to set up your objects to pass into flip.

[art3mis]

Thanks Ryan. Well, one thing I think needs to be covered more is 'best practices' in typical scenarios, especially in regards to  suggested starting settings for flip or pyro sims at both small tabletop or macro scale and large 'ocean' size scale.

For instance in another thread someone suggested that water should always have a feedback value of 1.0, yet in any test I've done that value is too high. I assume because it's dependent on many other factors.

Also, having a hard time understanding density as used in Houdini FLIP sims and when, why you would change it.

Yet another is Scale Time on AutoDop and its relation ship with Timescale on the Flipsolver.

Yet another is properly simulating things like surface tension in macro sims, though I understand that will be more or less 'built-in' to H16.

 

 

[rbowden]

I can agree that more 'presets' would be beneficial for anyone new coming into Houdini when it comes to flip.

- For the feedback value, it depends, again, what you are trying to do. This is straight from the help: To make particle fluids affect the motion of objects they collide with, turn up Feedback Scale on the FLIP solver’s Volume Motion. So if you are trying to make something float, then I would turn it to a value of 1. See the attached file for an example. In the file, I have the feedback scale set to 1 but if you set it to 0, you will see the sphere just drop through the tank.

- For changing density on the flip object, it depends on what you are trying to sim? You going for water? Keep it at the default. You looking to do a different type of material, you would change it. For example, if I wanted to simulate Tar, I would change the density setting to 1153. After that I would start messing with viscosity. Where did I get that 1153 number? From here:

http://www.simetric.co.uk/si_materials.htm

Here are a couple of more sites that could be useful also:

http://www.simetric.co.uk/si_liquids.htm

http://www.engineeringtoolbox.com/density-solids-d_1265.html

- As for Scale Time vs Time Scale, again it depends on what you are trying to do. I have only ever messed with the two parameters when doing slow motion type stuff. I have used Scale Time when doing pyro simulations (a long time ago) and Time Scale when doing flip stuff. Scale Time is a global time for the dop network. What that means is that if you have flip and RBD in the same network, they both are going to be influenced if you turn the Scale Time down to 0.5. In the same attached file I have made another dop network that illustrates what I am talking about. If you change the Scale Time on the AutoDop_Time_Stuff down to 0.5, you will see that the sphere touches the water at frame 22. If you have it at 1, it touches the water at around 12. This works out great because you are globally slowing down everything. So that means that the sphere going into the water is slowed down also. What happens if you want to go from fast to slow (like bullet time effect)? This is where one of the downfalls of the Scale Time parameter comes in. You can't key that value in a good way so that is when you would go back in and do the timescale on the flip solver instead. The problem with this is that the flip may be retimed but, the sphere falling down is actually still falling in realtime. In this case you would need to actually modify your source sphere and half the velocity on it basically so it will match up again. Kind of a pain actually and if you could get away with it, it would probably be better to retime it after sim honestly. For doing that, check out this thread:

https://www.sidefx.com/forum/topic/25640/

- Won't even bother going into surface tension. It isn't impossible to do in flip but, it does take alot to get it looking right. The simplest way would be to use the gas tension solver but, it can be painfully slow. You could take a look at what Alejandro did in his tool to see if you can retrofit it to do what you would like:

http://www.gridmarkets.com/alejandro-echeverry.html

If not then yeah, wait to see what 16 has to offer. Hopefully that cleared up some confusion for you. I didn't proofread this as I did this in a hurry so if something doesn't make sense, let me know.

 

fluid_examples_RB.hip