PDA

View Full Version : Multiple Interacting Liquids


HollyWoodland
09-12-2006, 09:37 PM
Has anyone tried implementing any of Ron Fedkiw's methods:

http://graphics.stanford.edu/~fedkiw/

As soon as I finish work (2.5 weeks to go til my resignation kicks in) I plan on trying out this one on Multiple Interacting Liquids:

http://graphics.stanford.edu/~fedkiw/papers/stanford2006-02.pdf

of course there will be a lot of work before i can even start implementing this method itself but it'll keep me busy for a while

I was really wondering if anyone else has tried this?
did you get good results?
do you have any tips on what to do / not to do?
or even any other methods that you think would work better than this?

craigd
09-25-2006, 06:42 PM
Yeah I tried the 'Animation and Rendering of Complex Water Surfaces' a few years ago. I was really happy with the result at the time (glass of water/juice filling up). Obviously my implementation of it was nowhere near as good as theirs, but I learned a lot along the way. I haven't seen any other papers that promise better results than the Stanford/ILM ones, and there are plenty of them too, it helps having the same ideas explained over in slightly different ways.

It's helpful to read the Stam 'Stable Fluids' paper and the Foster/Fedkiw 'Practical Animation of Liquids' first I think. The papers on the particle-level-set method in the top section of Fedkiw's publications, and the yellow level set book really helped too (worth the money). I found it easier doing the particle level set in 2D first, trying to duplicate the rotating zalesak's disk example he mentions. (It's amazing the difference it makes in preserving the area/volume).

I'd love to try out the octree one, and the multiple liquids too. Someday, when I get time to read through it all....

I guess you must also have seen this by now :

http://www.flowlines.info/gallery01.html

HollyWoodland
09-27-2006, 06:24 AM
It's helpful to read the Stam 'Stable Fluids' paper and the Foster/Fedkiw 'Practical Animation of Liquids' first I think. The papers on the particle-level-set method in the top section of Fedkiw's publications, and the yellow level set book really helped too (worth the money). [/QUOTE]

thanks for the tips - i think i read the papers a year or 2 ago - i'll have to dig them our again
I haven't read the book though - is it available from amazon? i'll have a look later - my last day at work is friday so i'm looking forward to getting stuck into something new (after a couple of weeks watching tv :bounce: )

[QUOTE=craigd]
I guess you must also have seen this by now :

http://www.flowlines.info/gallery01.html

actually i hadn't seen it - their R&D reel is awesome!!!

do you think they used a similar method to those above? it's given me some good ideas for my demo reel :) - assuming i manage to build something that works well that is!


have you tried any fire sims - thats something else i'm really keen to get working

HollyWoodland
09-27-2006, 06:24 AM
hmm my last reply got a bit messed up - half of my reply is inside the quote! interesting...

craigd
09-27-2006, 07:19 PM
yeah it's on amazon, just search for fedkiw.

i'm not sure what methods Scanline use, probably something similar though.

i haven't tried fire/smoke, just a liquid.

cheston
09-30-2006, 11:10 AM
Hmm, slightly off-topic, but I've been using Stam's technique for NS (from his 'Stable Fluids' paper that was mentioned) and have just found out it was patented by Alias! What's the deal with this and can I still use it? :sad:

neuromancer1978
09-30-2006, 08:43 PM
Hmm, slightly off-topic, but I've been using Stam's technique for NS (from his 'Stable Fluids' paper that was mentioned) and have just found out it was patented by Alias! What's the deal with this and can I still use it? :sad:

Patent law is something I am not familiar with, but as far as I know you can use it inhouse I think, you just can't sell it or even offer it to the public for free. For instance, look at BMRT. It was free (not open source) and after Larry Gritz left Pixar, he got sued for patent infringment and pulled BMRT off the net. But say if you were to use this technology for in house purposes and have no intention to sell or distribute - I think you wouldn't get into legal trouble.

But do not quote me on this! If I were you I would look into the matter.

cheston
10-01-2006, 10:36 AM
Ok, cheers.

That's what I was hoping, i.e ok for non-commercial use or similar. Not sure if there's any other way around it, i.e using another way to make the solver stable enough, but don't really want to go into that too much! :D

HollyWoodland
10-03-2006, 11:17 PM
ok Fedkiw's yellow book arrived today - seriously hardcore! that's any hope of sleep gone out the window now :)

nurcc
10-17-2006, 08:00 PM
I thought this would be an interesting link for you.. it's basically implementation advice for fluids. It tries to boil down the notation used in the various papers and books into something that's more straightforward for implementation.

It covers Navier-Stokes, a MAC grid, sources and sinks, fluid-solid interaction, variable and high-viscosity fluids, viscoelastic fluids, and smoke. It's unfinished.. it starts getting into level sets, but stops at fast marching methods.

http://poseidon.cs.byu.edu/~cline/fluidFlowForTheRestOfUs.pdf

HollyWoodland
10-18-2006, 11:21 AM
fantastic - thanks Tom!!

I've read a lot of the yellow book now - very hardcore! and plan to start reqs/design next week.

On a side note:
Just been looking at NextLimit (RealFlow) and 3D Aliens (Glu3D) - anyone know what algorithms they are using?
If they were UK based I would definitely be flooding their inbox with my CV (also Fusion CI Studios is doing really interesting work on fluids!) anyone know of any similar UK companies?

djwarder
10-18-2006, 03:32 PM
Hi

Both Realflow & Glu3d use SPH for their fluids, but not sure of any more specifics. I think a grid based solver is better myself, and from what I've heard the Lattice-Boltzmann stuff is coming into fashion. Having said that there's lots of info on traditional Navier-Stokes solvers ...

djwarder
10-23-2006, 04:08 PM
By the way, any Lattice-Boltzmann peeps out there?

neuromancer1978
10-23-2006, 11:09 PM
By the way, any Lattice-Boltzmann peeps out there?

Well since Blender has a LBM fluid solver, I do tend to play with it often. It is very impressive, but as with any fluid sim, it takes forever to do complicated and detailed runs.

djwarder
10-24-2006, 09:25 AM
Yep, the Blender fluid solver does look good, and should really have a play around with it myself. Really wanna start coding some fluids work of my own and think the LBM is the best way to go. I think I mentioned the patent issue with Stam's solver before, so LBM is a safer option in that respect, plus its faster and better than conventional solvers in some respects.

For me, I really am struggling to get my head round level sets and implicit functions, which are crutial to this area of work, so need to read up if I can. Have had a brief look at the Osher/Fedkiw book, but too heavy for me at the mo! Need to start at the basics of implicit surfaces I think and work up. Any suggestions?

CGTalk Moderation
10-24-2006, 09:25 AM
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.