realistic beer foam

[omere]

i've transferred the settings from your scene and checked the log. there are some parameters that was changed but i can't find them in the new LiquidSim.
this is the log:
keep_vxyz 0 -> 1
zmr 0 -> 24
spf 10 -> 4
spflimit0 2 -> 8
velvort 1 -> 0
vorticity 0.3 -> 0
cooling 0.05 -> 0
lqsharpness 0.5 -> 1
lqsurft 0 -> 0.2
drying 10000 -> 0.01
lqprt 0 -> 1
fbirth 60 -> 100
fbthres 50 -> 100
fsize 0.1 -> 0.215
frise 10 -> 30
fmage 1 -> 0
spfreefall 0 -> 0.5
spsplit 0.5 -> 0
spstick 0.5 -> 0
limitby 1 -> 0
prealloc_mem 1 -> 0
adapt_s 0.02 -> 0.1

i can't fins the following:
keep_vxyz
zmr
spflimit
velvort
vorticity
lqsharpness
lqsurft
lqprt
fmage
spstick
limitby
prealloc_mem
adapt_s

[Ivaylo Katev]

keep_vxyz - keep the grid velocity (output panel)
zmr - used for debugging, it exports different internal grid info into the smoke channel, so it can be visualized, for example how many particles are deleted in each cell, the pressure , etc.
spflimit - not used
velvort - not used
vorticity - not used
cooling - not used
lqsharpness - not used
lqsurft - in the dynamics panel, surface tension
drying - in the dynamics panel
lqprt - in the output panel , liquid particles
fbirth - in the foam panel, rate
fbthres in the foam panel, vel. threshold
fsize - in the foam panel, size
frise - in the foam panel, rise speed
fmage - in the foam panel, export - > age
spfreefall - in the splash panel, "by free fly"
spsplit - in the splash panel, "split to mist"
spstick in the splash panel, chaining
limitby 1 -> 0 - the adaptive grid limit , not used
prealloc_mem 1 -> 0 - not used
adapt_s 0.02 -> 0.1 - not used

[omere]

thanks.
i'm starting to get some good results now. but i have a lot of holes even with the cutter.
is it because i need to increase the grid size? i'im already in 3,300,000 cells

[Ivaylo Katev]

the cutter can't help here, in the old core the liquid-geometry interface was cube shaped, that's why the cutter was absolutely necessary, but the flip surface is flat.
sometimes the particles are just forming randomly holes, that is unavoidable during the intensive liquid movement, they disappear when the liquid settle. from user side you can resolve this using push modifier over the glass during the rendering, we can try to provide some automatic solution but can't promise. i suppose the shortest way is just to include the wetmap particles when the grid content is built, they are stuck on the walls and no holes can exist

[omere]

i wasn't aware of the wetmap. i will try and post the result.
so i don't need the cutter anymore?
thanks!

[Ivaylo Katev]

yes, in the most cases you will need no cutter. but it can be helpful sometimes, for example to hide a piece of the simulation that is considered wrong, or for artistic reasons
i was thinking to use wetmap in the grid representation, but even without this ability, the wetmap still can resolve the issue. there is an option of the wetmap, "sticky effect", it can prevent holes, i was using it for the beach scene, otherwise the force lifts up the water and there are very annoying cavities on the bottom.

[omere]

wetmap now is wetting? under dynamics?
and i need to increase the sticky effect?

[Ivaylo Katev]

yes, my mistake, wetting, not wetmap.
set the sticky effect it to 1. no need to export the particles, if you need them for the rendering

[omere]

it's looking good but i'm trying to reduce simulation time now. at some point after about 50 frames i'm getting to 1 hour per frame.
what is the purpose of the "Scene scale" parameter under Dynamics? and can it solve my problem of a too small of a scene?

[Ivaylo Katev]

yes, the scene scale can help, the scene scale plays important role for the spf requirements, you can increase it and decrease the spf.
however, i think the main reason is the huge bubble count. as i remember, your bubbles are very small, and this requires very big count to fill the volume. you can try to use bubble size variation with certain amount of bigger bubbles.
the most users are not expecting big change if one per 100 bubbles is about 4 times bigger, but you will be surprised from the result. 4 times bigger bubble occupies 64 times bigger volume, therefore this 1% of 4 times bigger bubbles will increase the volume by 60% . beside the performance effect, the bigger bubbles make the foam more natural looking, because the real foam has them too.

[omere]

i'm encountering a very long render time. eventually i manged to reduce it to the VRay light "cast sahdow" option. if i turn the shadow off then the render time become reasonable. here is the link to the scene:
http://we.tl/kYtOxWoMBX
maybe i miss something?

also, i've noticed that every time that i save a file, Phoenix will create a folder with the name "Untitled_Phoenix2_frames" even if i didn't used Phoenix.

[Ivaylo Katev]

the folder issue is already fixed but still not in the nightly builds, about the render times, will see the scene later, but if your bubbles are small enough, better to use point mode instead bubbles or cellular.

[Ivaylo Katev]

well, my observations until now
- you are using the slowest rendering mode, cellular. this is only for very closeup views. in fact, it's not important here, just for info.
- the light has 8 subdivisions, this means 256 rays per lighting point. this is needed only when the light source is very close to the object and the shadows are very soft.
- and the most important - the foam always produces "infinite noise", seen from the image sampler, and it forces tha maximum subdivisions, your subdivisions are set to 8, this is 256 times slower than fixed sampler. for good result 3 is enough
here is the result with these changes
with cellular/ bubbles (there is no difference in the view), 4 min and 22 sec

with point mode, 27 sec

[Paul Oblomov]

If only you could add support for per object subdivs for the foam the problem will go away, imo.

[omere]

i managed to get a good result. the only problem that remain is the foam head. it is going crazy and not stable.
why is that?

i'm adding here the scene for you to check:
http://we.tl/XVTEELQUl7
I started simulting from frame -350 so to have a full glass in frame 65