I’m playing since several days with liquids, and finally it’s difficult to have a clear understanding of relationship between these parameters.
I’m sure this have already been answered, but i found that increasing grid size to solve issues… gives even more issues ! (patterns in flowing fluid)
let’s start with a simple basis :
let’s say i have a good solution with 1M cells, quality 20, SPF 2.0 and surface level 0.5
if i double the grid size, how do i must change the other parameters to have similar result ?
Quite simple, but not so evident to figure it by myself after hours of testing.
Would be nice to have simple rules here, at least to have an idea to the direction to go when i double the resoution.
My goal is to understand, of course, and why not create a simple script to manage these parameters when raising grid res.
let’s say i have a good solution with 1M cells, quality 20, SPF 2.0 and surface level 0.5
if i double the grid size, how do i must change the other parameters to have similar result ?
i suppose under “grid size” you understand the grid resolution, i.e. change it to 2M cells. increasing the grid resolution in the most cases may cause artifacts that look like small waves. the only way to kill them is to increase the SPF, no need to increase it twice, 25% is enough, but to avoid the non integer value better is to use 3. usually i keep the spf value the lower possible, and even i allow these waves to appear, because they look nod bad, unless they are small.
but better attach a sample scene, let see what you get and how can we improve it. 1M is very low resolution and i wonder that you are able to get instability at all, the only reason that i can imagine is a very low grid size, about few mm
Hello Ivaylo, and many thanks for your answer. Helps a lot !
yes in Grid size i would say grid resolution (number of grid cells). 1M was a virtual case easy to understand for discussion. I finally finish often between 20 and 40M, depends of scene.
But, as you said, increasing grid resolution leads to little waves, that’s exactly my problem !
I didn’t knew if in this case i had to raise mainly spf, or quality, and so on.
Now it’s very clear ! The big thing id SPF. i’m trying right now the 25% rule
I also figured that raising the surface level to 0.7 or 0.8 increases the “little waves” problem. I figured that when using high res (30-40M cells) i could lower this value a bit to 0.4 or 0.3 to mask the effect, keeping enough details due to the hi-res of the grid. Is it correct/logical way to go ?
Thanks again for your help, i’m now ready to simulate the flowing water requested by Glorybound on other thread
the surface level is a rendering option, it does not change the artifacts, only can hide them. in my oppinion the surface level must be always 0.5 and should be changed in very rare cases, only as emergency solution.
keep in mind that the “25% rule” is valid only for doubling your cell count. the exact rule is thet you have to change the spf with the same ratio as the linear resolution change, and because the volume increases with power of 3, the exact value for doubling the resolution is 1.2599210498948731647672106072782
of course you are not obligated to use the exact value, and i would say it’s not recommended to use it, better keep integer spf value than use the exact scale factor. actually there is no “right” spf value, my rule is simple - i’m looking for the lower spf that does the job. this is for two reasons, first it simulates faster with lower spf, second it produces more details. of course if the spf is too low, the simulation starts to “imagine” details, and these artifact waves are exactly this.
my liquid tend to just disappear after a small fall. only surface level 0.3 or lower make it exist again. At 0.5 it looks like the liquid just vanish when hitting the surface after the fall. weird. I’m sure i’m missing something important here. nothing special, 20M cells, smooth conservation, 25 quality, SPF 5, classic method for advection. Vorticity 0.01, sharpness 0.1, only wetting checked.
I also figured that it is safer to clear the cache after many tries, right ?
hmmm… looks like it didn’t liked vorticity and/or sharpness set to 0.0. Better now, more consistent with vorticity 0.01 and sharpness 0.1.
But perhaps i’m wrong, and it’s related to quality ?
ok after testing, raising sharpness correct the problem, cool !
Sorry for these comments, i’m learning and posting at same time.
But i must say i already do love Phoenix
the vorticity plays no role in the liquid sims, but sharpness=0 (without strong surface option) means actually no liquid surface is supported and the behavior is like of a cold gas.
