Cell Size vs. Particle Sampling

[Felix]

Hello,

I am not sure about the connection between particle sampling and cell size.
As far as I understand, the paticle sampling is used to achieve higher particle counts without increasing the cell size of a domain. So if I want to have more particles, I would inrease the sampling instead of lowering the cell size. But where would be the point of doing that, if there is almost no difference in simulation times? The only benefit I can tell is the file size of the calculated field (*.gfc), which remains at the same size while inreasing the sampling.



Thanks
Felix

[LuisMiguel]

Hello Felix,

The difference between cell size and particle sampling is simple. Each cell has 8 particles inside. If you set a low value for the cell size, you are increasing the number of cells and there will be more particles because there are more cells for the hybrido domain. But each cell will have 8 particles inside independently the number of cells you set for the simulation. On the other hand, core sampling will increase the number of particles for each cell. By default, each cell has 8 particles cause the core particles sampling parameter is 2 at cube. So, if you set 4 in that parameter, you are indicating that there will have 64 particles inside for each cell.

So, setting low values for cell size will adjust the fluid in a better way besides adding more particles to the scene. It means, for instance, if you have 3 cells for your hybrido domain you'll have 24 particles with a cell size of 0.3. Now, if you set the cell size to 0.15, there will be 48 particles in the scene because you are doubling the number of cells. In the other case, if you have 3 cells for your hybrido domain, you'll have 24 particles, but if you increase the core particle sampling to 4, there will be 3 cell in the scene, but now, with 64 particles each cell, it means, 256 particles in the scene. The advantage increasing the core particle sampling value is you'll have more particles and the memory won't be consumed in excess, but it is not useful if you need the fluid adjusted perfectly with the geometry. In order to do that, you'll need to set a low value for the cell size parameter.

I hope it helps.

luisM.

[Felix]

Hello Miguel,

Thanks for your comprehensive explanation. I understand it perfectly and I am sorry for asking further, but given my example above (where I ran two simulations: one with cell size 0.6 and one with 0.3. Both have the same amount of particles because I incresed the emitter sampling while simulating the larger cells), both simulation times where almost equal. If the main adavantage it less memory consumption, shouldn't be the result a faster simulation? Or are you just talking about RAM and since in my case it only used about 20% it won't show any difference? I just played a little bit with the min/max particle sampling in the domain and there is definately just an additional number of particles without more computation..

[LuisMiguel]

Hello Felix,

Yeah, I was talking about ram memory. Depending on the scene, the 0.3 and 0.6 are big cell size values and maybe you won't be able to take advantage adding more particles when increasing the core particle sampling. Try to run a simulation with at least 10 millions of particles, modifying the cell size, and do the same modifying the core particle sampling instead.

luisM.

[Felix]

Hi Miguel, thanks for the clarification, it helps a lot.

Best regards
Felix