only some realflow fun

[tom schuelke]

Hi had some time to play with realflow.

Its easy to get interisting results. But its horrible to get exactly a look , your client wants...... grrr.


Tom

http://www.tom-schuelke.de/forum/waters-smal.m1v

[Candelero]

great results, bravo. Like the camera angles a lot.

Best regards,
nikki Candelero

[vlado]

It's beautiful

Best regards,
Vlado

[ene.xis]

yeah!

they look amazing!

inspirational stuff!

[DaForce]

Very classy!! Well done.

[Morbid Angel]

well done. However in my opinion realflow has a flaw, where when two liquids intersect, they do not collide with eachother, thus produsing no splash, where one would expect.

[cpnichols]

well done. However in my opinion realflow has a flaw, where when two liquids intersect, they do not collide with eachother, thus produsing no splash, where one would expect.

It is probably because those voxels are to small... the smaller the voxels the more likely you will see a splash... but the sim time... ouch....

[Morbid Angel]

chris, perhaps you are right. Imo the whole approach in RF is incorrect;
where you cannot assume that 1 particle represents 1 drop, as rf does. Liquid as other matter is composed of molecules, which can be broken into smallest fragments. I know Im talking light years into the future here, but thats why there are issues of this sort.

[cpnichols]

chris, perhaps you are right. Imo the whole approach in RF is incorrect;
where you cannot assume that 1 particle represents 1 drop, as rf does. Liquid as other matter is composed of molecules, which can be broken into smallest fragments. I know Im talking light years into the future here, but thats why there are issues of this sort.

never actually used realfow... how doe is to dot it? particals? voxels? level sets?

[jacksc02]

ooooo thats really nice, what were the sim times like?

[priad]

Nice test!! alway cool to see RF stuff

[Morbid Angel]

chris, realflow works on the principal of particles that represent the liquid, which work with dynamics and etc. Once the particle solution has been reached, where you got the desired animation result, you can then build a mesh based on the particles. Think of it the same way as reactor builds convexity hull over the mesh, only with some controls. In RF, each particle has a radius. If your mesh (which is actual fluid we see in the end result) falls within this radius RF will generate the mesh. Sometimes particles can have smaller radius, i.e. 1 particle wont generate a mesh but several particles clamped together will. The mesh has connectivity like meta surface. So that the particles near eachother blend.
This method has its downfalls. For one, in realflow its hard to make particles from one emitter to collide with particles from another emitter. Secondly, particle radius is an approximation of the liquid's shape, not an exact form. So you cannot split the mesh infinitly. Finally, if you think about a real life liquid spread on the surface like a table, it has a tendency to me flat on the bottom (collision side) and have a round shape on the top. This is not possible with RF, as the mesh will be build in a spherical form no matter what. Unless you use a special cutter to terminate the mesh which will produce a hole in it.
In the end, simple simulation, to semi complex simulations are possible. But RF will give out in about 3.5 million particles. Which isnt a whole lot...
the end

[cpnichols]

chris, realflow works on the principal of particles that represent the liquid, which work with dynamics and etc. Once the particle solution has been reached, where you got the desired animation result, you can then build a mesh based on the particles. Think of it the same way as reactor builds convexity hull over the mesh, only with some controls. In RF, each particle has a radius. If your mesh (which is actual fluid we see in the end result) falls within this radius RF will generate the mesh. Sometimes particles can have smaller radius, i.e. 1 particle wont generate a mesh but several particles clamped together will. The mesh has connectivity like meta surface. So that the particles near eachother blend.
This method has its downfalls. For one, in realflow its hard to make particles from one emitter to collide with particles from another emitter. Secondly, particle radius is an approximation of the liquid's shape, not an exact form. So you cannot split the mesh infinitly. Finally, if you think about a real life liquid spread on the surface like a table, it has a tendency to me flat on the bottom (collision side) and have a round shape on the top. This is not possible with RF, as the mesh will be build in a spherical form no matter what. Unless you use a special cutter to terminate the mesh which will produce a hole in it.
In the end, simple simulation, to semi complex simulations are possible. But RF will give out in about 3.5 million particles. Which isnt a whole lot...
the end

That seems a bit primitive. I know that Aura uses Voxels, the fluid sim at DD was also voxel based that then created level sets. Basically what you are telling me is that RF uses metaballs?

[joconnell]

Yup it does. RF is one of the least intuitive, unstable, unfinished programs I've ever used. My life gets significantly less pleasant every time I have to do a job with rf. Anything aside from a standard water is really irritating to try and get, the support is horrible and very few people seem to have a good understanding of how to get the results you want...

[DaForce]

and to get RF to fill up a glass or something without poking thru you need a SH*T load of particles so you can set the radius small enough that it doesnt just stick thru the glass and making it look like crap.

I think somewhere i read that they recommend using thicker objects so you can use less particles, and a bigger radius therefor faster, and the particle mesh sticking into the main mesh isnt noticed as much cause the main object is thicker.....

Its just a pain in the ass to use most of the times.