Lets talk about Mantaflow

(esimacio) #401

regarding graphicall download speed, it took me about 15 sec to download 75Mb

(LordOdin) #402

Well graphicall downloads crash a lot. A few people on FB complained about it today I sent them my own link since graphicall isnt usable a lot of the time.

The downloads terminate in a really strange way makes it feel sketchy lol

(loran) #403


Here’s a test I ve done. Can’t figure out why drops don’t appear on contact under the fall. This is where white water sould be. Particle behaviors are not natural, I don’t know if a realistic effect can be done with this system. Maybe I am doing it wrong but for now, I only see one only good result with mantaflow particles. I posted it up there.

(loran) #404

What I learned : Adaptativ stepping have to be off to avoid crash on up to 90 divisions Resolution.
Fluid diffusion panel Size have no impact. I don’t know how to scale the scene for a glass of wine or a water fall
I don’t know hove to avoid stepping in particles on fast motion. The sampling subframes seem to create crash but doesnt resolve stepping.

(zebus3d) #405

Waves test.
I do not understand why water never falls down it after goes up the hill.

(esimacio) #406

hmm have you tried any of the forces?

(juang3d) #407

What build are you using?

in the latest versions you have an scale value, you should have adaptive stepping on if you want good results with collisions in general, maybe try the latest versions of Mantaflow to check if the problem is solved.

Ofuscado: The water may be not going down because the resolution is too low and it is “traped”, it may be also because the volume of water you have is to big and it indeed occupies this volume, but I think it’s a resolution problem, anyways I have a similar scene to do some tests soon, I’ll check the same thing :slight_smile:

(yuzupei) #408

Hi,did mantaflow support spilt simulate job to multi cpu on network ?

(juang3d) #409

Nope, at least not right now, I´m not sure about that in the future, it could be awesome, but I donñt even know if the actual mantaflow system can do that.


(Dito) #410

Hi LordOdin,

your renderings looks fantastic!

A small tutorial maybe?

(kumayuki) #411

Sim time:11 hours 45 minutes

I used the 2017-10-24 brunch.

When simulating the same scene with the updated branch, the result got worse.

(yuzupei) #412

11-29 version the domain can’t get the right initial value. but when change the collision’s distance value,not change other value,the correct result will get back.

(loran) #413

I am using the last one ( 2017-11-29)
Scale value (size of the domain in meters) doesn’t affect the simulation in the different test I ve done.
adaptative stepping crash every simulation over resolution 90

(loran) #414

@kumayuki Thank you for sharing your settings! I will try this :slight_smile:

(zeauro) #415

I confirm that.
You have to do that before baking anything or velocities for first frame are wrong (too high for most of particles) and then, each following frame will be wrong.

(zebus3d) #416

Now with 200 in resolution.

(Razorblade) #417

Its kinda strange the resolution effect, i see what happens.
However at the deep water there is no visible effect of these ‘blocks’.
Might it be possible to have adaptive block size ?. (not sure if thats possible) ??

(zebus3d) #418

I think that when you can adjust the resolution of the collider individually this will be fixed.

(juang3d) #419

I think there is going to be something like an “antialiasing” for this kind of things, but it’s not there yet.


(zeauro) #420

Some FLIP particles are probably stopping their movement. Try to modify obstacle characteristics.