New Liquid Solver: When to raise spf and when to raise cell count.

[Ihno]

Hi,
after some weeks of trying the flip thing i'm wondering which one of this is responceble for what?

Seems that raising the spf is leading to more accurate simulations and also raises the particle count.
The count of grid cells on the other hand seems to affect the collision quality. Higher count = lesser particles will run through walls and so on.
Is that right?

And how to set it up for quicker tests? Low amount of grid cells?
Low spf seems to let the liquid suddenly explode some times especially in combination with surface tension.
So i assume that a low spf isn't suitable for quick tests.
Is that right?

And last but not least have you any recommendation or rule of thumb how high it should be set for the final sim?
With the the old solver it was recommended to set the cell count >50m at least I read that somewhere.
Is that still the case?

[Ivaylo Katev]

yes, for fast test lower res is better than low spf.
sings that you need higher spf is segmented stream, randomly flying particles, any instabilities or non decaying movements

about the resolution, the requirements for production are the same, at least 50M cells

[Ihno]

So if i'm happy with what going on in my tests I just have to lower the cell size till it reaches 50M?
Probably it depends on how much of the simulator is on screen.
I'll have to get at least 50M if the whole simulator is in the camera view and set it higher if i want to show just a part of it, right?
Also i think it depends on how much of the cells are filled with liquid. A 50M grid wouldn't be enough if I just fill 2M of it with liquid.
What kind of artefacts would be related to a understated grid?

[Ivaylo Katev]

the cell size and the spf are related, in many cases when you increase the resolution, this causes lack of spf and you have to increase it too. but this happens not always, first you increase the grid size, and if you observe some indicators for low spf (segments, random splash, oscillations) than you must increase the spf too. the general rule is - keep the particles movement to be less than 5-10 grid units per step.

[Ihno]

Ok, thanks alot.
Would than be nice to have an option in the preview rollout to show if particles move more than 5-10 grid units per step I think.

[Ivaylo Katev]

you can use the white color of the preview, set it to value 5 x spf x framerate, and if you have big count of pure white particles, this means may be the spf is too low. but don't forget, there is no strong rule, sometimes only 2-3 particles are violating the spf rule, and increasing the spf only for them is useless. in general, do not trust this rule more than your eyes. if you have no segmentation or flying particles, forget the rule.
but if you have such problems, you can use the white speed method to test the spf criterion, sometimes not the spf is to blame

[Ihno]

Haha I should start to formulate such things as question.
How often I sayed something like "it would be nice if..." just to find out its already in there.
Thanks for clearing that up.