cleaning process in slow motion

[Jassir]

Hello,

I was asked to create an animation with water in slow motion. The water is used to clean a piece of metal.
The animation starts with a closeup of the metal, the camera is moving slowly. A water jet enters the screen rather fast from the side. The water jet cleans the metal part in real time for some frames and changes in slow motion while the camera circles the droplets and water structures. After approximately 180 degrees the time switches back to real time, the water jet is gone and the remaining water flows of.

I thought this should be possible using phoenix. Unfortunately I'm having some problems. My first approach was to use a cylinder as source and animate it. Using this method provided me with very little control and I had some issues to get a steady jet.
Now I'm trying to use particle flow with a deflector and vray metaballs for the jet and as source for phoenix. I think I should have mor control using this way but it is still hard to achieve good behavior.

Is this way a good choice? Am I missing an important trick? Should I use a completely different approach? Can I animate the timescale parameter for the slow motion effect.

I'am grateful for any advice.

[Ivaylo Katev]

better use nightly build and flips solver, it is particle based, no need to use pf at all.

[Jassir]

Thanks for the reply. Sounds interesting. I never used a nightly build before and I can't seem to find a general explanation what I have to do. Could you point me in the right direction.
I found the Link to the phoenix nighlies in the phoenix forum but when I log in with the account we used to buy phoenix, I only have the option to download Vray for blender.
Is there something I should have done differently? Does it work with the standard Vray 3.0 build?

[Ivaylo Katev]

you have to write a mail to the support, explain why you need the nightly (or just use link to this thread) they will explain you what to do.
the nightly works with official 3.0 vray, no problem. just keep in mind that not all nightly builds are stable.

[Jassir]

Thanks for the tip with the nightlies. From what I saw so far I would say the flip solver is a really nice feature. But is it correct that in order to use it I only have to change the Advection method to "FLIP (Particle based)" or are there other settings that need to be changed as well?

Another thing I have to solve in this project is the time remapping in phoenix. As far as I understand it I have to do it with the resimulation toolset (described here: http://docs.chaosgroup.com/display/P...ime-scale,etc..)
But there is also the timescale setting in the dynamics rollout. Do I have to combine these somehow? Or are those for different workflows?
What do I do if I only want to re-scale parts of my animation. Say I want to start with normal speed, slow down for about 2 seconds and go back to normal speed?

Third issue is the rendertime with splashes. I have arround 10.000 splash particles in my testscene (one material, simple geometry) and the rendertime increases from 7 seconds to 1 minute and 19 seconds (480x270px). Is that normal or did I forget something important? Settings for splashes are left at default with the exeption of "geometry mode". I turned that on, otherwise vray crashed during the calculation of the irradiance map.

[Ivaylo Katev]

yes, to use it you have to switch the advection method. the other settings are dependent on the scene, in your case due to the high velocity of the liquid, you have to use high spf.
you can't use resimulation with the flip solver, it is still not implemented, but you can animate the time scale parameter and this will slow down the simulation, in my opinion this is the best way.
about the render times -the reason for the slow speed is the GI usage. for example scenes with million of particles are rendering in 20 seconds but with no GI. if you need gi for the rest of the scene, you can use compositing and separate rendering of the particles and the geometry

[Jassir]

Thanks for your help. I tried to apply your tips to my test scene, but I encounter some strange issus.

Following these instructions from the phoenix documentation:
Time scale (MXS: timescale) - Time multiplier. Can be used for slow motion effects. When you change the time scale, in order to achieve the same look, you must change the Steps per frame accordingly - e.g. decreasing the Time Scale from 1.0 to 0.5, you must decrease the Steps per frame from 4 to 2 as well.

I started with an SPF value of 20 and a time scale value of 1. Than I dropped the SPF to 1 and the timescale to 0.05. After some Frames I returned to the starting values.

In the Result you see how the liquid, that is supposed to be water looks more like glue, "explodes" as soon as the values change. If I change the sharpness value sometimes the water flowed upwards as well.

Any idea what I did wrong?

Here are some of my settings:
Cell size: 0,6cm
x: 70,2cm
y: 70,2cm
z: 79,8cm
Total cells: 1 820 637


sharpness: 0,9
viscosity: 0
non-newtonian: 0
surface tension: 0
wetting: deselected
particle export: selected
velocity:selected
ID: selected
Age: selected


Conservation: Direct smooth
Quality: 30
Advection: FLIP

Here is the testvideo:
test02.zip
And the max file:
SlowmoTest02.zip

Any help is much appreciated.

[Ivaylo Katev]

the statement in the help is about the situations when you need exactly the same result in two separate simulations, it does not apply to changing the timescale/spf during the simulation. in few words, don't change the spf, only the time scale. just as additional info - we are abandoning the concept that the spf can be changed during the simulation. the old versions of phoenix are able to work with changing spf, but this leads to unacceptable complications

[Jassir]

Thank you for your responses. I tried it with an fixed SPF in my testscene but it didn't make things a lot better. Maybe something else was wrong. So I created a new testscene. The fixed SPF value of 6 worked a lot better than what I had before. Unfortunately I still have some issues. I attached a video again for better understanding.
I like how I have these small bubbles of water at the beginning. It gives an impression of a strong impact which is important. But when the timescale value drops, I lose these bubbles and the water becomes very simple. It looks more like a soft flow than a strong jet.
Any idea how I could fix that?

Video:
test03.zip

[Svetlin.Nikolov]

Unfortunately, animating the simulation timescale would indeed produce changes in behavior like the one you have. The only solution in this case is time bend resimulation (a.k.a. re-time), which has to be implemented for the FLIP advection. Something else you can try is to simulate with fixed spf and fixed time scale, also with particle velocity exported, and then animate the play speed form the Input panel and use Velocity blend method. This method will eventually be replaced by the time bend resim, but it might help in this situation.

[Jassir]

Thanks for your help. I'm trying to go with this approach now. But I ran into another problem. No mater what, as soon as I try to play back the simulated animation I always get an error message that states "Could not blend particles! Please make sure they are exported with an 'ID' channel!".
And I can't seem to find this missing ID channel. I selected following ID channels to be exported
- Liquids
- Particle Export
-ID
- Splash/Mist
- Export
- ID

I also selected all other channels concerning particles (age, velocity).

There are no particles produced in the source object and there is no foam.
I have set an absolute path to the cache files.
If I scrub through the animation I see the preview. I just can play it, which is necessary to animate the play speed.
And if I try to render it, it works perfectly.

Strange thing is that I receive this error message if I deselect all particles as well.

And I didn't even start to animate the play speed yet.

[Svetlin.Nikolov]

Interesting, we'll check this, stay tuned.

You may check "don't show" on the warning message so it won't bother you (it will be remembered forever for this scene only after you save the file).

Meanwhile, if you change the play speed, does it slow down properly?

[Svetlin.Nikolov]

Okay, so I will disappoint you - velocity interpolation is also not yet functional with the FLIP advection. You will either have to revert to the classic advection method, which will allow for time bending effects, or wait at least a couple of weeks until blending for the FLIP method is implemented. Also, I have to remind you that if you decide to try with classic advection, you have to set the steps per frame to at least 4.

[Jassir]

Unfortunately not so well. I looks a bit pulsating, somewhat back and forth.
I set the Play Speed to 0,2 (fix) because it wouldn't let me set a key.

Here is my short test:
slowmo4test.zip

edit:
Sorry, I didn`t see your post there. I will go back to classic than.

[Jassir]

I' trying to go with the classic advection method now. But I'm still having problems. First of all, I still get this error message "Could not blend particles! Please make ...". Although I'm working with classic advection now. This makes it impossible to use the time bend controls provided by the input panel.

Then I tried to go back to the standard version of phoenix (not nightlies). But I had to realize that the resimulation of liquids in order to bend time isn't possible there.

Any idea how I could avoid this error and use the timebend controls of the nightlies builds?