Fluid Simulation Research Demos

Posted on by

water_oil.jpgPablo over at Max Underground (the Max version of BlenderNation) posted about some updated fluid simulations on Ron Fedkiw’s Stanford site.

The most interesting are probably the Multiple Interacting Liquid videos that are definitely a must see.  Though we don’t have the ability to siumulate multiple viscosities like this in Blender (does any commercial application though?), it’s still worth sharing with everyone simply because fluid simulation in Blender is a big “selling point” and we can only hope research like this gets incorporated into Blender one day.

Be Sociable, Share!

    Twitter



    Please support us

    • Matt Henley

      Interesting… I havnt seen it being used for 3d graphics but my company uses it for simulation of performance of 3-phase oil/gas/water separators. We use a package called Fluent ( http://www.fluent.com/ ) which is extremely computer intensive as well as expensive (a typical setup would be leased for on the order of US$30-40K/yr ). There is open source code (check on openflower) as well. One resource is http://www.cfd-online.com/

      Matt

    • Hoehrer

      O My God! This page is a gold mine with cool stuff in it.

      It’s sad that it will need _a lot_ of time until we get something compareable (especially the smoke&flame simulations .. which have been reported before soemwhere … and the changing of mesh topology during simulation) in blender :-/

      Werner

    • http://www.blendernation.com Eugene

      @Matt
      That is some serious fluid sim that Fluent has.

      @Hoehrer
      Does any 3D app have changing of mesh topo during simulation?

    • Hoehrer

      @Eugene:
      Not that I know of … that looks like some serious work-intense procedure (code wise as well as manually) though so I’m not sure how the guys from Stanford did the videos/simulation.
      It would be a great bonus point for blender though … since something like this would prove really useful in movie-animation/special effects.

      Werner

    • QEDqubit

      jpl, jet propulsion laboratory has created -in coop with others- a simulation program that’s VERY interesting, called”paraview”
      just google 4 “paraview-2.4.4-win32″ and press the ‘i feel lucky” button, there’s only ONE hit leadin you to
      http://www.paraview.org/HTML/Download.html

      this is a first attempt to an all round simulation engine, and the demo shows airflow around a jet fighter, fluid in a box and lots of things I DON’T understand.
      REAL BRAINIACS ONLY
      or well… I’m OFFICIALLY stoopid so, lot’s of WILL dig this beautiful software, and even may connect it to blender verse later on…

    • QEDqubit

      oh.. it’s open source and thus available on linux offcourse..

    • keir

      @QEDqubit

      ParaView is a generic scientific visualization suite; not a fluid simulator. You can use ParaView to visualize the output of a fluid simulator; however it does not by itself simulate fluids (or anything).

    • http://niel.seyanim.com madman

      fckng awesome!!!
      Great site! I download all.

    • http://www.konrad-haenel.de konrad8ha

      …makes you wonder how long some of these simulations took to solve.

      …and what kind of creative freedom things like that would allow.

      ahhh, dreaming on…

    • brecht

      The latest videos from Nils (the developer of Blender’s fluids) are also very impressive:
      http://www.ntoken.com/p_fluid.html
      Check especially the “Detail-Preserving Fluid Control” video, being able to actually control the shape and animation of a fluid like that in Blender would be great. Although this depends of course on what is allowed to be open sourced and how much time Nils has.

    • http://gustavthemushroom.deviantart.com Tim

      I think Microsoft used something similar to this for their game studio logo/flashy anim.

    • Big Fan

      wow that “Detail-Preserving Fluid Control” video is great.
      The fluid horse and figure were awesome! :o)
      I wonder if Blender will get something like that in the future? – maybe too heavy going though for your average pc…

    • toontje

      On another note, Hugues Hoppe’s page has some 2006 updates too. And check out the old realtime fur demo.Just amazing. It runs very smooth on my old PC.

    • http://www.lessequal.com bobef

      I’ve seen this site before comming from here http://www-etud.iro.umontreal.ca/~clavetsi/physicsingraphics-details.html . Check it out. A lot of cool stuff plus many of these guys even provide the formulas for their algorithms…

    • Andrew

      Hmmm, I posted an article relating to this some time ago on blener.org http://www.blender.org/forum/viewtopic.php?t=6257&start=0. To this day I’m still slightly confused as to how fluid simulation has made its way into blender whereas simulation of gaseous phenomana such as smoke have been largely ignored.

      The maths and coding of this is far beyond me, but I would love to see an implementation of Jos Stam’s Stable Fluids algorithm in future releases of Blender. Here’s to hoping.

    • http://pencilproductions.googlepages.com clinton

      My life is better because of this.

    • epat

      ooh – I’ll definately give this paper a read when I get the time :), I like this sort of stuff quite a bit. Contrary to popular belief though, the main problem with decent fluid simulation in blender and especially with smoke and flame simulation is that lack of proper rendering facilities, meaning that anyone who gives it a go would probably need to write a LOT of new stuff into the blender rendering engine to get it to look even partially respectable afterwards anyway, the maths behind this sort of stuff is hardly a problem in comparison to the sheer amount of coding in other areas of blender to get it working properly and nicely too… anyway, enough moaning – You never know, I might be able to make a python script to do it – though I’m not sure how I’d get all the fluid data itself to remain between frames if it was a scriptlink – I could write it to a text file and parse it next frame I suppose but it would be terribly slow – mind you, in python… – my main problem is that my C programming skills leave somewhat to be desired ;) and as I was saying about rendering capabilities anyway… -thank’s for the link though – I hadn’t noticed these new updates were on here, this might have just made my day :)

      -epat

      btw: about changing the mesh topology during simulation – basicly, with this type of fluid simulation (eulerian – i.e. grid based methods), the mesh is recreated from scratch every simulation step anyway so it’s not as bad as you think…

    • bearX

      ohh! that is great!

    • http://home.tiscali.nl/dotblend/ peter

      http://home.tiscali.nl/dotblend/BLENDERFLUID.html

      created in blender, rendered in blender…
      different fluids combined..

    • http://www.blendernation.com Eugene

      @Peter
      Please tell me you have the .blend available to share!

    • http://home.tiscali.nl/dotblend/BLENDERFLUID.html peter

      @eugene

      http://home.tiscali.nl/dotblend/BLENDERFLUID.html
      the blend file is available

    • freakydude

      what codecs do the movies use?

    • http://www.blendernation.com Eugene (etr9j)

      @freakydude
      They are DivX

    • bludragon

      I’ve ran a few CFD demos on my Linux box (using ParaView and a CFD engine for Linux called OpenFOAM). The good news: its really cool. The bad news: it takes my computer a very long time to do the analysis (not unlike doing a lot of fancy animation in Blender).

    • jccorreu

      OK so this question is for people who know the code behind Blender.
      What parts of the code are currently responsible for the fluid simulation stuff. And what parts would have to be modified to get more of what people are ohhing and ahhing over?