Hi everybody, I'm trying to write a fluid simulator based on the following paper: physbam.stanford.edu/~fedkiw/papers/stanford2001-02.pdf I am using a level set and store one phi-value at the center of each cell. However, to evolve the level set I need to define the phi-value also at the WALL-cells. I have a couple of questions on the implementation of this: 1. Which value do I assign to the wall cells? (As my fluid region stores positive distance values to the closest surface I assume they have to be negative, but which value?) 2. At the interface between WALL cells and the fluid region, which (positive) phi-value do I assign the fluid? As it has to be a signed distance function, are these values +1 at the WALL interface or actually the closest distance to the next AIR cell? My main problem is that for a breaking dam simulation the water seems to stick to the wall, but if I replace the level set with particles it all works well, which means that my level-set code must be wrong... Any answers to above questions or hints on how to solve this problem are mostly welcome! Thanks, Icebraker

im sorry,
i cant answer your questions, and id even like to ask you some...


what is the main reason to choose "level set" and not any other method?

what are the other comparative methods to achieve the same thing?

are your cells fixed in space and size during simulation?

why not use particle version? is project about visualization or analysis?

how can you so easily switch between cell and particle dynamics, would not initial parameters to set up simulation physics be of completely different type for these two?

Hi,

OK, I'll try to answer your questions:
1. level sets allow the quick extraction of a smooth surface at very low cost compared to particles.
2. yep, my cells are fixed in space and size
3. Well, as I said, the extraction of a smooth closed surface with particles is not trivial and can get quite expensive computationally if the simulation volume is large
4. No, I used marker particles which just get moved along the velocity field. But as I said, it is difficult to extract a smooth continuous surface.

I still hope that someone can answer my initial questions though...

thank you,

some more, if you don't mind...

Quote:
Well, as I said, the extraction of a smooth closed surface with particles is not trivial and can get quite expensive computationally...

what is the difference between:
-"extraction of a smooth closed surface with particles"
and
-"let particle interaction produce the surface by "filling in" the geometry"

if i understand correctly, the data/result you are after is not what happens during the simulation, but some description of some surface after water has settled?

what "surface" are you after? the surface of the "lake"? how much will flood spread, how much will it be absorbed by soil? water level after flood?


cheers