rgb simulation time?

[Ihno]

Hi,
it's a bit difficult to say for sure. But can it be that rgb raises the simulation time in liquid sims alot?
I've been testing a sim, especially the outgoing velocity, about 200(edit:100) frames after the sim started.
I had about 10mins per frame sim time.
After I was happy I went back to the beginning of the simulation and activated rgb.
I'm using an animated noise map to emit it.
Now the sim is roughly where I tested it before and I've got a simulation time of 1h and 57min.
I changed some other things like the cell size and the spf a little bit.
So I expected the time to double or even triple but two hours seems ridiculous.
I've a grid with 18m cells and 3.1m liquid particles. No foam or other particle types.
Im using the latest stable in max 2017.

[Svetlin.Nikolov]

Wow, this really seems like too much. Would it be possible to send over the Phoenix log from the slow run? It will answer which parts of the sim take the most time.

Cheers!

[Ihno]

You mean the slog files in the settings folder?
For some reason they seem too old. From the beginning of the month which is where I started to play with this sim. But who knows... might be that windows forgot to update the change time.
Here is the settings folder, the scene and the latest cache file.
Thank you for beeing so responsive.

[Svetlin.Nikolov]

Hey,

No luck reproducing the issue here - the sim is running at 7 secs per frame. Probably it's caused by some of the external assets - I'm missing all the jpgs, the vrscene and the xref-ed max file, so all of these are suspected.

You can check several stuff:
- Does removing the RGB map from the source make any difference.
- Does removing any of these assets above make any difference.
- I meant the Phoenix log file which you can access from Simulator -> Preferences - it will also supply some pretty good info on what's going on.

Cheers

[Ihno]

ah ok here is the log.
I'll review the scene in meantime. Thanks!

[Svetlin.Nikolov]

Hey,

So it seems like the Droplet Breakup + Droplet Radius might be the cause for the simulation slowdown. Increasing the Droplet radius can cause the simulation to slow down a lot. Do you remember if you increased it before it got slower? Try lowering it to 2 or 3 just for testing if this helps the sim speed.

Also, there is a warning in the log file that the 'Emitter' object is not a closed geometry. This could cause all kinds of issues in the simulation, so watch out for that as well.

Hope this helps pinpoint the issue!

[Ihno]

So it seems like the Droplet Breakup + Droplet Radius might be the cause for the simulation slowdown. Increasing the Droplet radius can cause the simulation to slow down a lot.

Thats good to know!

Do you remember if you increased it before it got slower? Try lowering it to 2 or 3 just for testing if this helps the sim speed.

Might be, I dont know.
After I was happy with the end of the emitting I setted the beginning. So it could be that I raised the droplet size.

Also, there is a warning in the log file that the 'Emitter' object is not a closed geometry. This could cause all kinds of issues in the simulation, so watch out for that as well.

Fixed, thanks for the pointer!

Seems to be much faster now (~4min@f1061) with a droplet radius of 2 and a closed emitter. Thanks again.
Would a change of the Droplet size be visible in the simulation? Like when changing the spf.
Resim that from scratch wouldn't take very long now though.
But could I use 5 for the beginning and when the liquid flow reached the maximum velocity, animate that down to two without creating a big visible jump or something like that?

[Svetlin.Nikolov]

Hey,

Glad this helped! Animated Droplet Radius will very likely be visible in case the Surface Tension and Droplet Breakup parameters are high. The transition will not be abrupt, but if the simulation is in such a state that round droplets have already formed and are clearly visible, animating the radius to produce larger droplets will probably cause them to visibly reshuffle, but you should be able to get away with a change such as maybe 1.0-1.5 over a prolonged period, and if the simulation is chaotic during the transition, this will serve to cover up the effect even better

Cheers!