Nils Thürey, who brought fluid simulation to Blender, has done it again! He's co-author of a paper titled 'Wavelet Turbulence for Fluid Simulation', an algorithm which creates stunning smoke effects. Even though the algorithm is not implemented in Blender yet, the code is already available. Check out their site for some stunning videos!
Abstract from the paper:
We present a novel wavelet method for the simulation of fluids at high spatial resolution. The algorithm enables large- and small-scale detail to be edited separately, allowing high-resolution detail to be added as a post-processing step. Instead of solving the Navier-Stokes equations over a highly refined mesh, we use the wavelet decomposition of a low-resolution simulation to determine the location and energy characteristics of missing high-frequency components. We then synthesize these missing components using a novel incompressible turbulence function, and provide a method to maintain the temporal coherence of the resulting structures. There is no linear system to solve, so the method parallelizes trivially and requires only a few auxiliary arrays. The method guarantees that the new frequencies will not interfere with existing frequencies, allowing animators to set up a low resolution simulation quickly and later add details without changing the overall fluid motion.
Ton Roosendaal comments:
"Best news is that the code has been released as gpl! Blender developer Daniel Genrich is very interested in looking at integrating it, not to mention I would love to see it as Durian vfx!"
Link:
- Wavelet Turbulence for Fluid Simulation. Includes full paper + stunning videos/
62 Comments
OMG OMG, will this code be available into blender 2.47 ?
Ok.....now im excited...
WOW
Wow ! I hope that this feature will be included in Blender one day.
Congratulations to the coders. This tool is really amazing !
@Owen: "will this code be available into blender 2.47 ?"
Read carefully ;-) "Blender developer Daniel Genrich is very interested in looking at integrating it"
It may take a while before you can actually use this.
I am also stunned!!! We own you so much Nils Thürey. Thank you!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
@Bart > Hmm.....I see :-|
Don't forget that Farsthary is also working on smoke, see http://farsthary.wordpress.com
Exciting times ahead... :)
What is the difference between this and farsthary's True Volumetrics ( http://farsthary.wordpress.com )?
Of course not. But what is showcased is really promising :)
(this was a response to Owen)
@Guran : Farsthary's volumetrics make it possible to fake smoke using textures, whereas this will actually simulate smoke the way fluids are simulated, in a very realistic way (which would probably be a lot slower...)
guran: Farshary's method is a "classic" one: quick et realistic fluid simulaiton like Maya's one, ...this is perfect for little flames and smokes.
It method enables realistic mid-sized flames and smoke with little cpu-power.. ..but still not sure about the original paper.
But he is aware of this paper, simply he did the implementation in early august and Nils'paper code source was just published about a week ago.... Otherwise he would have already put the code inside blender..
Daniel/Farshary could make a kick-ass team! ;)
I haven't a clue as to what all the technical stuff means, but this sure sounds impressive!
WOW!
Great !The video demostrations looks Amazing!
..better have both methodes integrated!
(Farstharys Volumetrics (Smoke Fire) and this from this Amazing Paper)
because farstharys Smoke looks little different ...so we have more Inovation and artistic free making smoke and fire...with many different methodes....etc.!
Yeah!
Farsthary will resume school (=daily internet connection) at the end of the month, so a branch should quickly be set up to optimize the volumetrics required for these effects.
Great things going on. I love Blender Community, it's the most powerfull and most capable community in the world of internet. Go Blenderheads, GO!!! :)))
Is it live or is it Blender?
Damm man, this is really powerful and great. Keep on coding guys, please.
Hugh: no, Farsthary code is also based on fluid physic: it shares the first steps of the Nils' paper:
1) fluid simulation based on Navier-Stokes equation (real fluid physics) on a coarse grid to consume little cpu-power
2) extrapolation and smoothing on a higher definition grid to avoid grid artifact
Nils: 3) information about energy density in the grid are used as basis to generate a detailled grid with vortices (vortices are here to dissipate the energy), the displacement of each vortex being given by the grid from 2)
Hope this is a approximate resume
Nice research... would be great if we can get this into Blender
Neat!
wow
amazing look! i´m asking myself if it would it lead to an advanced fluid-sim, too...actually, there seem to be a lot of problems with the obstacle-identification (slip-values and so on...) and the handling is more like trial-and-error then a controlled simulation...
Yay, I can't wait!
@yoyoyoy: +1
I was thinking along the same line. It seems a no brainer to me that this would also be applicable for the regular fluid sim. So basically, your massive fluid sim wouldn't look like mud creeping along in the city streets, but brutal splashes everywhere. That was the main thing that was off in Ice Age 2 when the dam burst. The water looked like textured hair gel.
you see, this is why i love open source software, it only takes one person out of all 6 billion of us on this little rock we call earth to think of doing something and voila, there it is.
great!
*dreaiming of candy floss*
So it is possible to make something like ink pured in water, or a sediment transporting river floating into another river or ocean? Red and whithe Nile or the big Amazonas.
OMFG!! OMFG!!! Have you guys seen the latest changes on Farsthary site? This is the the best smoke simulation I have seen in Blender! Check out the soot/ smoke, I can almost smell the fuel burning.
Yeah, I hope they work together on this because Farsthary is a good ways down this road already. Great news!
This smells good indeed...
I hope these improvements can be unified and interact with the rigid and softbodies of the physics engine!
wow!!!! I am already thinking of all the things I be using it for lol YAM!
I wonder if the algorithm can be added to the current particles system. I think it would be really cool to have objects move about like smoke.
Hi Daniel/Nils/Farthary,
as smoke requires volumetric, both smoke generation and smoke rendering should be as quick as possible,
perhaps that you can take some points of this paper to optimize the smoke rendering:
http://www.kunzhou.net/2008/smoke.pdf
In this paper, the idea is to separate the smoke in 2 contributions through maths technics:
* the low frequency smok = main one
* the high frequency smoke = little variations due to vortices:
Under certain lightning conditions, low frequencies can be realistically rendered in a less rigorous way, leading to great speed boost.
This paper makes uses of a preprocessing step to separate the 2 kinds of smoke, but the smoke generated through Nils ' algorithm is by definition separated in the 2 contributions above, so this can achieve an overall algorithm (smoke generation + rendering) which is quicker!
Thank for your contributions!:)
OMG!!!
I wolud like to see this code implemented in the blender 2.47 or Blender 2.5 for VFX in the Durian project
Excellent news. Looking forward to seeing this in Blender.
this is just sick madness ;)
I hope it will make it into Blender soon !!!
Tonight we dine in battle!
This smoke is most impressive simulation I have seen yet- fantastic results.
Aww dammit! I just bought lightwave and the dynamite plugin because blender had none of this stuff! ($1200).
Stuff from lightwave still looks better though (but I imagine that'll change real soon).
No pun intended but, holy smokes! This is so exciting it's giving me the hiccups.
Definitely the timing of this and Farsthary's work couldn't be better during the run-up to Durian. Great things in store for Blender!
GraphicsKid: You bought it? I downloaded the 30 day trial and a crack.
He hasn't stated that he will develop this for Blender, but still, this is awesome nonetheless!
Looks pretty cool. Still, there's something weird about it, like those "ridges" in the plume shaft. Overall a little too water-ish. Nitpicking on research code, I know, but there's things that still need to be done here to get it to behave more like real smoke. Perhaps simply more nonlinear deflections. The "ridges" stick out in my mind, tho. Hopefully this will be worked out beautifully.
Amazing.
Tynach, I wouldn't talk about piracy here.
GraphicsKid, you did good. By buying the software you have a legitimate right to use it. Those who like to break copyright to get stuff for free run the risk of being sued or worse.
That said, I think this smoke thing is insanely awesome.
Awesome stuff
*Waves hands in excitement*
This was some stuff I really needed back when doing accident-simulations!
Ohhhhhh so cool! that made me stay up way past my bed time.
Damn... that's "smokin'" hot! Can't wait for it.
Thanks, dudes!
great news!!
Howitzer: I never used it, and I don't have it anymore. So, I'm safe. Actually, I never even really installed it either. Just used Blender.
vidrazor: Are you referring to the videos, or the images? I know that I thought the same things, until I saw the videos, and watched a real cloud of smoke. Sometimes you don't realize how real something looks until you see the same effect in real life.
Looks long way from finished, but hey. it's Nils. It'll improve. btw, can you make it light the evniroment to, >.> we really need that. I can continue my vertex copy based area lights, but it's not as good. or is lightcuts bringing that to us? nyway, it'd also make it a strong tool competing with FUMEFX and Afterburn. nice work on farthsays to. it's great for when you don't need a massive long calc fire effect, just a good candle or something.
Holy Smoke Batman!
That is an incredible simulation algorithum.. It looks so real.
I watched these vids just a couple weeks ago! Wow, it would be pretty amazing to see this in Blender, really smart stuff(haha, can you tell I'm not very algorithm savvy?).
smoke is nice but i rather have caustics, I also see they still havent learned that ".MOV IS BAD " ;P
I never thought that fluid sims could be used for smoke. This is something that I could really use!
I've got mixed feeling about those videos.
At first my jaw dropped, my eyes were peeled to the screen and I started drooling.
But then I got kinda sad, because I know sometimes new features like this never gets implented. Kinda like how a new cancer remedy hits news pages and then we never hear of it again.
I guess you could argue that I should get off my arse and join the developing team to help out. And you'd be right there. So I'll shut up :P
I really am quite amazed at the intellect of programmers,
especially the ones that work on Blender. They give us virtual
tools with which to attempt to simulate and re-create the real world
with more creativity.
Thank you all Blender developers, especially Nils Thürey and Daniel Genrich!
It looks very nice... BUT!
- The wavelet turbulence makes all the turbulence look the same.. no variation, just the same uniform turbulence look all over the simulation. I hope its very tweakable when/if implemented
- It won't look good anyway without a decent volumetric rendering engine that handles shading and shadowing properly. (When is such a thing coming to blender?)
Can't stop watching it. Amazing!
i m trying to compile the open source in my mac but i cant... can anyone help me?