Fluid Simulation Alternatives


(AustinC) #1

Does anyone know of any alternatives to the Blender liquid simulator? The internal one is weak for large scale and for higher viscosity, so an alternative for those would be most desirable. Most importantly, how would one such external simulators be integrated into a Blender workflow? I have seen a few promising simulators, but I have not seen anything regarding how to use it for a Blender scene.

I am aware that there are ways to fake many things, but I could ask questions in a different thread regarding specific effects that I wish to achieve. For now, I just want to know what other options I have.

Thank you.


(El Director) #2

I would be curious to know this too as I’m needing to do some large scale tsunami sims for an upcoming movie.


(fitz301) #3

I too am trying to get a movie made that requires large scale bodies of water as it’s based on the true story of a shipwreck, so I’m curious as well.


(fitz301) #4

I’m also wondering if there are any plans on the drawing board for enhancing Blender’s fluid simulation and ocean modifier?


(AustinC) #5

I’ve been playing with the particle fluids in Blender. It seems to scale much better, although you need a LOT of particles for detail, which is very normal. And, to get the mesh smooth, I had to use metaballs twice the size of the particles (due to the influence distance being larger than the particle). So, you’d certainly need a massive amount of RAM and processing power for a large simulation. Using larger particles, the fluid simulation, or the ocean modifier for bulk fluid and using fine particle fluid for smaller droplets and details seems like a decent approach. The wave dynamic paint with the ocean modifier looks reasonable as a starting point for adding particles.

I am still working with the particle fluid to see if it will work for what I need, but it is a pain that setting an initial volume of fluid seems nearly impossible. To lack this fundamental feature, perhaps it is just for effects to supplement the internal fluid simulator.


(fitz301) #6

Do you have any example videos or screenshots of this?

I never thought to use metaballs for fluids, I would like to get an idea of how you combine the two. I’m not really that fluent with Blender’s particle system yet.


(pingking23) #7

there are several commercial softwares availible:
realflow
houdini
phoenix by chaos group


(AustinC) #8

I did not save any videos or screenshots, but I will make a quick tutorial video on rendering particle fluids with metaballs. I will post the link here when I finish the video.

None of those seem to support Blender. And, while Houdini seems to have a reasonable price for small studios ($200/yr/seat), Realflow is quite expensive at $1100 minimum per seat. Phoenix is in the middle at $830. And, again, none of them support Blender. It looks like they all work through plugins.


(fitz301) #9

I would appreciate that - thank you.

I’ve been looking into other alternatives as well. Have you seen these sites?


(fitz301) #10

Did you try molecular?


(AustinC) #11

I have seen Fluid v3, but not the others. As an engineer, the tools that actually aim for accurate CFD are extremely appealing to me in that regard. But, I see some issues for using them with Blender. The simulations are all external, and I do not see that any of the geometry exported from Blender to the simulator can even be animated. Moreover, I do not see a large advantage of using them over the build in particle fluid. The VTK importer for a meshed fluid seems good (better than metaballs), but that is about it. As far as I know, you can use millions of particles to get the same effect in Blender, but all methods lack the sense of scale from fluid details. They allow much more detail than the internal fluid simulator, but millions of particles only gets you so far.

The answer seems to be having fluid that emits smaller particles or finer particle fluid. The fluid would need to emit these smaller particles based upon the velocity and the density. I have no idea how this could be accomplished in Blender.

The other issue is viscosity. I do not see any high viscosity tests for any of those simulators.

I have seen it, but I never tried it because I only saw crude simulations on YouTube with fairly large particles. It seems to handle higher viscosity, initial volumes of particles, and even mixing of materials. This may actually work for my needs, but I still want to look into how to use finer particles with it. I will have to look into those later.


(pingking23) #12

what is you aim with the fluids?
why do you need them in Blender?
houdini and realflow can import a lot of geometry formats and also export the meshed fluids back which can be imported into blender


(julperado) #13

Inside Blender you could use Molecular and CubeSurfer (both from pyroevil), but those are still very slow solutions since are not modifiers.

Or, If you choose to use any other program just export the final sim as alembic and import it into Blender.


(AustinC) #14

I am working on technical pipelines and assets for some animations that I have scripted. They are small animations, and I don’t expect any kind of revenue. I don’t want to mention specific situations because the purpose of this thread is to know what simulation options are available.

I don’t need them in Blender. But, they would need to interact with animated objects.

Molecular or particle fluids in general will be very difficult to make look convincing without splash effects with smaller particles. I’m not sure how to do that in Blender, but I think Realflow and Houdini can do it.

I did not see any alembic export options for anything mentioned in this thread yet. Realflow and Houdini, for instance, seem to work through plugins. And, the open source options seem to only use VTK outputs.


(AustinC) #15

After playing with it for many hours, metaballs are not a good way to do it. And, the method is rather complicated. The viewport vs render mode isn’t working, so the viewport has to show the rendered mode (with the metaballs), which is very slow. Moreover, it takes a very long time to build the scene. And, you lose too much resolution in the fluid by having to make the metaballs large enough. And, how large the metaballs need to be changes based upon your simulation settings. So, I would say to just use CubeSurfer instead.

Overall, I’m giving up on the particle fluid system inside Blender. The model is not based upon physics, so making it look plausible is nearly impossible. It is not good for anything other than splashes and such to compliment Blender’s fluid simulator. I don’t have the time to look at Realflow or Houdini in depth, but Realflow seems like it would work the best.
I looked closer at Houdini, and it seems that the only version with a fluid simulator that could work costs $4500.

DesignSPHysics looks good for simulations with very simple dynamics, but I have no idea how the workflow of using it with Blender would work. For fluid that interacts with complex animations in Blender, I just need to make the internal fluid simulator work. I will need to fake effects wherever possible and edit scripts to avoid fluid effects that I can’t fake, that I can’t handle with the Blender fluid simulator, or that I can’t handle with DesignSPHysics. $1200 is just too much for what I’d use it for.


(pingking23) #16

houdini indie has the all the tools from the “big” version, its only restricted to the HD output and the revenue of you/your company.
houdini is not the easiest software to learn, but its probably the most flexible for any kind of simulations.
you can import animations as alembic, do your simulation and then decide to render directly in houdini or export the mesh of the fluid

you can also test houdini total free of cost with the apprentice version, there are some restrictions to the output but for testing and learning the best option


(AustinC) #17

I miss-read the output restrictions on the Indie version. That makes it more appealing. It certainly looks like the best option. Thank you.


(tobbew) #18

I also think that houdini indie is the wisest choice. There was a gsoc this year which involved fluids, you could check their videos if you haven’t already, but from the demo videos I don’t think it’s for large scale use. (Also likely not stable enough for production either.)