You're blocking ads, which pay for BlenderNation. Read about other ways to support us.

PhysBAM has been released!

61

It can be a hassle to try to mix different simulation methods in blender, so what do you think? Could this physics/simulation library be implemented in blender? Thoughts and comments below!

PhysBAM is a multiphysics simulation library, capable of simulating rigid & deformable bodies, compressible & incompressible fluids, coupled solids & fluids, coupled rigid & deformable solids, articulated rigid bodies & humans, fracture, fire, smoke, hair, cloth, muscles, as well as many other natural phenomena.

61 Comments

  1. [Facepalm] Arrgh. I'm all for making Blender the best it can be - but just when it's nearly stable, a whole new toolset gets proposed! I'm not convinced that this set is better than the one we've been waiting for a year and a half to get through Blender Beta. Let's get what we have NOW stable and released before we look to the future!

  2. Foam is something I'd really like to see in in Blender's fluid simulations. This video shows a really good example of it. The fire also looks really nice.

  3. I dont see why not. it's a separate lib it could be implmented as a GSoC and it would be nice in blender to be able to chose physics engine.

    some advantages is that you can run many diffrent sims in the same domains, they're striving to get it working in realtime … there's many advantages.

    blender maintainance could just be having interface classes towards the lib.

    it would be stupid not to try and integrate it , physbam has 13yrs of development longer than blender has as open source. :P and then we would have a physics lib in blender that's maintained and further developed by PhD's and students etc …

    I cant help feeling like libs such as elbeem etc has almost halted in the physics development.

  4. This is what i have been looking for! No need to get scripts to do it for you. I think this would be valuable for improving Blender.

  5. I don't see why not, especially considering Mango might need this. Some of those simulations look very accurate. I'm drooling over the water & burning fluid mix.

  6. JustinBarrett on

    I am impressed with the physical interpolations like fire on water, and the ocean breaking waves turning into mist....how fast is it, and blender is just now stabilizing.....if it could be converted to openCl and then implemented say 2012 GSoC....that would be fantastic.

  7. I agree that combining the different sims in blender is a bit of a hastle at this point, and I was particularly impressed by the surf waves crashing against the lighthouse tower in this video (something that will fry the brain of the most advanced CPU if attempted in blender now, and still won't look nearly as good); the man's face while talking was also very impressive (attention: makehuman team :-))
    However, the traditional management question of benefits vs. costs arise. First, as far as I know, blender is implemented in Python and C/C++ and is mutli-platform, whereas PhysBAM, it seems, is implemented in C/C++, exclusively on Linux, so there is an issue of porting and integrating into the existing code (or turning it into a platform-independent API that is implemented through dynamic libraries\plugins for the corresponding platforms: from the look at the PhysBAM website, it seems that it was designed with this approach in mind, but I haven't delved into the code itself and don't know what pitfalls are there). And another thing: I myself am a great fan of at least one tool--the aforementioned makehuman--that, while not being a part of blender, is closely coupled with it. I cannot give any more examples from the top of my head, but I'm sure there are other similar projects afoot. If a totally new physics engine is to be adopted by blender, how will it affect them? In other words, how much coordination is there, on the management level, between the "mainstream" blender and its "off-shoots?" Can all of them adapt it in an orderly, synchronised fashion to avoid breaking the artists' existing work?
    But I will say it again, if this library does become a part of blender, then blender will become a lot more powerful and a few things that are currently off-limits will become easily available (big waves, foam, muscle bulges when moving, and what not).
    So I would suggest putting it into the product backlog (to use SCRUM terminology) and see when the time is right to get on with it. As @Amusionist said, making what's already in place stable should be the priority.

  8. I agree to make Blender stable but what I love Most about Blender is the Massive amount of a variety of tools that even though I do not use them all, I may need one some day.

    This tool WILL BE USED because it makes simulations better and faster(I think).

    If anyone can implement this and Keep Blender Stable lets do it!

    Also how was this video made? What 3d software was used to make it?

    Thanks
    Scott

  9. Francesco D'Ambrosio on

    This project seems to be awesome and i love what i just saw.
    In my opinion, anyway, we just focus on finishin the 2.7 version of Blender to really work on something solid and stable, and THEN start working on others implementation.
    I would definitely love to see this physic engine integrated with Blender in the future anyway.

  10. Blender's physics are in need of an overhaul for Mango so why not give this one a try. Other 3D packages offer the user the choise over the physics sim via plugins, I think that's the way to go for Blender, too!

  11. This looks fantastic and some of the simulations are pro .. butI have to agree with some of the other folks. Get Blender stable and then add new features. Even though this would be great for some of my video projects ...

  12. It does seem powerful, but as others have said, aren't we a little late in the game to change direction now? I suppose if it could be implemented as an add-on, and not take any current developers away from getting 2.6 ready, it might be worth looking into by someone -- an add-on physics engine, sort of like an external renderer.

  13. Integrating this for Mango would be a huge amount of work, but nothing like what the team had to do to get Blender 2.5 ready as they (mostly) did with Sintel. The muscle, fire-combining-with-oil and large scale fluid demo's far surpass what is currently possible so this gets two big thumbs up for integration from me!

  14. PhysBAN have some really nice simulations, it would be great to have this implemented into Blender along with the current Physics engines we have.

    The wave effect was great, the oil on fire was awesome, and that bullet cracking down those glasses was really good too.

  15. Wow, Wow, Wow,
    Oh my god, the Cloth sim was blowing, and the waves were jaw dropping, and don't say anything about the muscle sim it made me dream of my future movie.
    No No No, this must be implemented the sooner possible, at least as a separate soft (like RealFlow of NextLimit) where we can export and import sims between blender and PysBam.

    But As always, Folks you must include in the video demos the PC specs (CPU used RAM used) and the render and simulation solve time, it's a must to compare the computer where this demo was done and our hardware (may be this demo needed a special computer config, that cannot be afforded by occasionally blender users, right!!??).

    I want to suggest an idea: Why not make all the blender developers switch to PhysBAM to make it implementable the quickest possible and then switch back to blender to make it stable with the team of PhysBAM involved too, like wise we concentrate the effort of the two teams on one project a time (we call this in army: FIRE CONCENTRATION).?

  16. Guys, you're forgetting how easy the solution is towards both keeping with current development and potentially adding PhysBAM to the line-up: make PhysBAM a plugin. That's one thing 3ds Max and Maya both do. Instead of changing the whole structure, just add it on as a plugin. Problem solved. We used to do that with Blander 2.4x, with YafaRay and MakeHuman. Why not do something like that again with 2.5x?

  17. I agree with Brian L. this would be perfect as an addon that uses blenders materials and such, you could even implement it to an addon to the current physics, but i dont know how well that would work out

  18. Denis Russkih on

    Can I ask, what software was used to render this video? PhysBAM - multiphysics simulation library. And renderer - ???

  19. The Masked Lurker on

    The possibilities for software like this are endless.

    Especially the softbodies which seem to stress the actual volume of the body being calculated when a deformation occurs. this feature alone is something that I've been praying for for a very long time.

    This would allow for more realistic muscle sims as well as more realistic fatty tissue in characters than the current soft body paradigm where the edges of a mesh are basically mathematical springs which flatten out unless you create massive internal geometry. Even then it won't allow that great a deformation without making it look too rigid.

    The fact that these various features seem to be able to interact in simulation is fantastic! Even if softbodies and fluid sims were able to interact in Blender as it currently is I would be very happy but this is a whole new ball of wax!!!

    This will eventually be added... it must, I don't care how or in what form but with what Nick Bishop and a few others are working on like Bmesh (and Farsthary? I hope his automatic tessellation feature eventually is able to be finished and added, although I understand why it wouldn't), I cannot wait to see what I can make with these new innovations.

  20. Why not, It have to be there. As i see blender is heading towards a great future and of course hope to see this greaaaaaaat physics in it.

  21. if you take a look at the PhysBam website (http://physbam.stanford.edu/) you'll notice that part of their project is a raytracer. Presumably they used that to produce the clips.
    I'm not sure that implementing this library is as a blender plug-in is as feasible as many people seem to think. Even the blender game-engine is not implemented as just a plug-in. There are several parts of it hard coded into blender's source. It might be possible to do it that way, but it would still be an immense amount of work to make it work as well as it should. It might even be better to just replace the current physics engine with this one in some future release. I don't know.

    That being said, I would absolutely love to see a physics engine as powerful and well supported (about 30 PhD students, about 10 undergraduates and masters, a stanford faculty member, and a few post doctorates).

  22. Wow. Nice looking stuff for sure.

    You could always use a workflow where you render these things in PhysBAM directly and then use compositing to put it all together. Sharing the models would be the main obstacle, which probably isn't so terribly hard.

  23. Idea:

    I think blender needs an integrated plugin library that you could launch inside blender. There would bee free open source plugins and paid plugins. Think of it as an app store, or Anrdoid market. Acually Android is a very good reference because it's also open source. That way blender would provide an all-around 3D platform for free, but the developers would have the opportunity to make some awesome plugins and earn something too. There would be much more plugins available and blender could be seen as an awesome 3D ecosystem that you can get free and expand as you wish, if you want more professional stuff.

    I think it's great that blender ships with a set of disabled plugins right now, like the alignment tool etc, but there's a tight control which plugins get into the releases. Also memory intensive plugins like texture libraries cannot be integrated in the main releases for obvious reasons. The 'plugin app store' would provide a very easy way to find also textures, materials etc in addition to plugins. That way the main releases could also be kept clean of any 'extra stuff', because it would be so easy to download them from the store.

  24. If you can implement it with a clear UI, that would be awesome... Clearly separating multi physics from the regular ones but so that you don't compromise and choose only between one and another. Interactions between engines is not needed nor really possible but I think simulating both would be nice like when you wish to use the regular cloth and smoke but wish to use multiplysics for the skin and breakable objects.

    So any one object could have shared controls panel section where options for "collision able", "stiffness", and so on... multiphysics section where you can enable different calculations, after that blocking the regular physics controls, or regular physics panel for the currently implemented physics with blocking multiphysics options. Also regular object physics that has to go through game engine are just not intuitive at all.

    If the options are scattered all around and some of the physics options are mixed, that would be just horrible. So clearly separating the two would be real nice. But I am looking forward multiphysics in blender. Having that option would be really REALLY nice.

    Best of wishes to whoever wish to implement this. :)

  25. whaaoo ..
    So so good

    Could this physics/simulation library be implemented in blender?
    oh yes :-D

    for me it's ok :lol: really beautifull the Ragdol and the Waves are so interresting :-)

  26. Not sure if anyone is aware of what PhysBAM has been used for in the past. PhysBAM was created by Rob Fedkiw who a computer scientist and is often recruited by visual effects houses (like ILM) to solve physics effects for them. This is what he does and his research papers have been shown at SIGGRAPH many times. I have known about PhysBAM for many many years (2003????) so to see it here was a shock and quite frankly exciting. So I figured I would offer some support for this and shed some light to the Blender community as to where it came from and why PhysBAM could be one of the best things to be added to Blender in some time.

    Rob's university website : http://physbam.stanford.edu/~fedkiw/

    For instance, with ILM, PhysBAM was used to create some of the ocean effects for the movie "The Perfect Storm", in fact there is a R+D pic on the site as a testimony of this. - http://physbam.stanford.edu/~fedkiw/papers/stanford2002-03.pdf

    Or the work done for the Terminatrix in "Terminator 3", this was done with the PhysBAM code base. - http://physbam.stanford.edu/~fedkiw/papers/stanford2003-04.pdf

    There are many other examples that can be listed, if you go through his publication list I am sure you will find more surprises on PhysBAM's use in the visual effects industry.

    In my opinion if this replaces all the other physics sims in Blender, the better. Imagine having the ability to do ALL physics with one library, not one for a cloth sim, another for smoke and a totally different one for fluids. It makes no sense to have different physics sim in Blender when this one could be added and solves ALL physics computations.

    SO I say bring it on!!!

  27. i agree bullet was nice but now we NEED physBam,
    it is the way of the future plus the code is there all thats left is for someone to integrate with blender and fix any bugs
    sounds easy but i know its not that easy
    regardless blender needs to move forward
    and if this is forward lets move forward, especially since autodesk already has features like fractures and others,
    it is only natural, in order to become a major physics player in the industrial cg community :)

  28. John Q Public on

    I would greatly appreciate it if the Blender developers would focus more on stability, basic functionality and interface design before they delve into creating more "deep" features like advanced simulation. There are a lot of really basic things that need to be fixed first: we don't even have a modeling system that supports N-sided faces, making modeling very difficult; the hotkey customization system isn't as developed or functional as it should be, and some of the defaults are extremely inconvenient - for example, switching between vertex, edge, and face selection modes cannot be changed from Ctrl+Tab+1/2/3, meaning that 3 keys must be pressed for a single mode change; the whole program is still quite crash-prone; and there are serious interface issues, such as the fact that there is no warning pop-up when you go to save over a file- I've repeatedly accidentally saved over files simply by clicking one too many times on "save as."

    These videos are all from tech demos for research papers from Stanford University. They've released some parts, but not all of the PhysBam library, mainly the low-level parts.
    http://physbam.stanford.edu/links/getcode.html
    They've released a source-code demo for the smoke simulator, but not liquid , cloth, rigid bodies, soft bodies, or anything else, as far as I can tell.

    http://physbam.stanford.edu/~mlentine/project.html#smoke
    http://physbam.stanford.edu/

    So it looks like implementing any of this into a usable form would be a massive undertaking, and not one that the Blender developer community can really afford to be embarking upon at the present moment.

  29. Ditto. Let's say it's for version 3.0 of blender. For one, I want to see Blender coming out of Beta. You have to admit Blender has nifty simulation features... Bear in mind: it's all make believe - in other words, the artist still has a say, and if it doesn't look good, tweak it some more, and let it crunch the numbers.
    Now, if you read the Stanford background and history
    http://physbam.stanford.edu/links/backhistdisclaimcopy.html your realize that they are (and apparently will be) constantly in flux. We need some stability. Their goals are probably divergent enough from ours that latching on / integrating their libraries may not be all that easy after all. Let's hope it is, but a measure of realism is in order.
    Still, I'm drooling. What a video!

  30. wow, this is some amazing stuff... it is amazing for sure, if we could use these tools even with an outside tool I would be on it

  31. Wow, this would be great in blender....It will definitely be used, and the thought that there is a chance that we can get those kinds of simulations in blender...it boggles the mind...

    I really want blender stable, but this is the type of thing that will bring new users and make a lot of old ones much happier :)

  32. There are all sources not just smoke. I'm using blender daily and die to see those features as soon as possible!

  33. Wow. I pray for the day this level of simulation is integrated into Blender. If I had a clue how to do it I would... Someone could make a lot of money making a plug-in implementing this improved physics. A bloody fortune I think.

Leave A Reply

To add a profile picture to your message, register your email address with Gravatar.com. To protect your email address, create an account on BlenderNation and log in when posting a message.