AWS Thinkbox Discussion Forums

Meshing Anisotropic - takes ages

Hey guys,

is there anything we have to know about the anisotropic meshing alg. ?
Because it’s really weird, it takes ages to complete the meshing.
And sometimes it stuck at 4.76 % and nothing happens.
When i switch to metaballs or something else, than it just took 5-7 secs!
Also with the default settings, i have changed nothing.

Totally weird.

I’m loading a 40 frames Bin file from Realflow ( 220MB per file)
But also when i just load a single frame with everything at default settings - nothing happens.
The Info Line ( left Bottom Corner ) is just blinking like hell ( 4.67%) and nothing happens anymore.
Maybe after 50 secs it gets to 4.68%.

cheers

naik

i also noticed anisotropic is slower than zhu bridson, but you have to be gentle with her! if it takes forever its because the settings are “not quite right”,

first thing i would do is lower the polygon count, then go from there, playing with the surface levels, etc.

Thanks for the info mate.

Actually i’m starting with the default settings and all meshing algos are doing “quite” well
expect for the ani. But it seems to be the radius -really strange.

because the default parameter is 0,5 i think ( very rough ) and with that it takes ages!
when i decrease it to 0,005 then it speeds up the meshing process…

gotta do some more tests.

It would be cool to hear some thoughts of the developers which meshing algo is best for
perhaps water / foam / splashes etc . Maybe there is a rough guideline for that.

cheers

naik

As you have found, Anisotropic meshing speed depends a lot on the particle radius. As a starting point, try setting the Radius to around the distance between adjacent particles. To help you visualize this, you could try switching to Union of Spheres while you are adjusting the radius.

We’d like to make our Anisotropic meshing implementation faster and easier to use than it is now.

I don’t think we’ve put together any guidelines, but it’s on our wish list. For water and splashes I’d recommend Zhu/Bridson or Anisotropic meshing. Union of Spheres and Metaballs tend to look more blobby.

cool,

thanks for the heads up Paul.
Appreciate it!

cheers

naik

In some cases metaballs with a lot of relax on top of it can produce nice results too.

Good point, this is definitely worth trying as well.

Privacy | Site terms | Cookie preferences