Check the article here:
Sorry but the article is PR department nonsens. Even in the paper adstract:
Meshes that conform well to changing boundaries and that focus computation in the visually important parts of the domain can be generated quickly and reliably using existing techniques.
There is no new “fluid” equations and this is just another approximation.
For the record here are some other points:
First of all we have been doing fluid dynamics with all the accuracy we can get since computers were computers. Most don’t care about apperance they care if the plane will fly of not. This means codes that conserve relavent properties.
Secondly any computer must approximate the fliud (there are 1 billion cubic millimeters in a square meter! you run out of ram). But thats OK it can still be accurate.
Thridly computer graphics does not need accuracy but good looks. The fluid pluigin in Mayer is not even close to accurate. It dumps all compressive flows so that there is only vorticity left. In fact the blender fluids is probably one of the closest to the true NS equations out there.
I’m reading the paper now. (I should be writing my own paper)
Delt0r: Just out of curiousity, can blender’s fluid system handle multiple liquids such as oil/water?
No not at this stage. But nt keeps adding things and his orginal work included air/water type simulations.
Oh i read the paper. Its is just a “moving” mesh method where the entire domain is remeshed each time step. Odd aproach really, the animations look cool tho. But they don’t claim to have new fluid equations in the paper. So Perhaps the PR was the reporter putting there spin on things.
1 Like
The smoke looks cool but the liquifying character doesn’t look like anything that Blender can’t do with a little more work by N_T with his current method.
It may be just another method of doing fluids overall.
I think you read the wrong paper, this article is about “Stable, Circulation-Preserving, Simplicial Fluids”:
http://www.geometry.caltech.edu/pubs.html
I find it actually quite an elegant method they propose, using the vorticity equations neatly put into discrete differential geometry. Of course it doesn’t allow new things really, but simply a faster way to do the same thing, like most graphics papers. This paper was already in the siggraph 2005 course notes, but only now published.
Yes you mite be right, I just followed all the links and only one lead to any articles.
However as far as fluid dynamics is concerened Its not new. They even say that in the introduction. Its new for visual fliud dynamics. My points above still hold. there is no new fluid equations, its still approxamate etc.
However it does look interesting.