I can’t see it anywhere actually.
Also, is manta flow automatically the engine for build sims in blender 2.9? My explosion looks as bad as the old blender fluid sims, so I wonder if I have some massive underlying issue.
Although you are using different settings for domain and fluids objects, you should not obtain a radically different simulation by using mantaflow.
That is not normal to see nothing in Viewport in Solid mode.
But if you played with Color Mapping ramp under Viewport Display panel, you may have tweak other settings under this panel.
In that case, it is possible to set a Thickness value that is ridiculously small to the point that smoke is no more noticeable but will become visible when color mapping is enabled.
Try to increase Thickness value.
For smoke invisible at render, problems can be different according to used render engine.
EEVEE is requiring pertinent Samples and clipping values.
Cycles does not have this problem. But domain’s material may not be correctly set.
Hey guys, quick question. Keep in mind that I’m complete goddamn noob with simulations in Mantaflow and I don’t fully grasp all the functions and proper use for them. I’m trying to punch a keyhole in smoke screen but I’m not getting any results. I’m setting up smoke sim, wait until it fills the desired space and then I’m trying to use Effectors - Colliders animating movement through said smoke but nothing is happening. If colliders are static they properly interact with smoke that curls nicely around the objects yet moving objects through the smoke yields no results whatsoever. What am I doing wrong?
I’m trying to recreate effect similar to one visible in Astartes - Part Three:
How would you go about creating this effect?
try to increase the sampling substeps of the effector
Doesn’t seem to be the issue - I was using max settings from the get go but it just doesn’t really change anything between 1 and 10. Also I’ve noticed that both Collision object and Effector: Collision object both interact with smoke however in different ways. Smoke coming in touch with static Collision object kinda wraps around it, while as soon as smoke touches static Effector it slingshots the smoke around the domain. Also was thinking about using force fields but it affects entire domain and doesn’t seem to be possible to scale it down? Even at 0.0001 strength it just screws up whatever is inside the domain like a hurricane 
I can’t wrap my head around smoke simulations in 2.8x+ (currently sitting on 2.90.1). Is there some decent, indepth resource showcasing different settings for Mantaflow smoke and proper usage of guides that I could look up to be enlightened?
Someone in YT comments suggested to set up object as both collider and soft body when animating but I didn’t have opportunity to try it out yet. That would be the weirdest fix. Well I don’t know if anything needs fixing it’s most likely me pressing random buttons and not getting anywhere… but I threw a lot of shit at the wall and nothing sticks…
Use the power factor to make a falloff of the force field
Here is a quick test scene with a moving obstacle and a turbulence with a falloff, all working fine (2.90)
TestSmokeCollision.blend (802.2 KB)
2020-10-26 14-56-35.mkv (467.8 KB)
A new APIC simulation method has been added (basic support) in 2.92
paper behind the method:
https://www.math.ucla.edu/~jteran/papers/JSSTS15.pdf
Video explainer:
12 Likes
I don’t understand the technical details, but the video seems exciting enough. Does this have significant implications for liquid simulations in Blender?
APIC is another method to simulate fluids (among many others), it tries to “correct” problems encountered in other methods (in this case FLIP) but it would create other issues in one form or another.
TL;DR if you use APIC you will get smooth and stable results compared to the noisy unstable FLIP results, but you may lose something else like preserving energy.
1 Like
You can exclude demos about sand and gelatinous cubes.
Liquids are the only case where Blender will use that.
If you look at video, for liquids, result of FLIP ratio at 0.95 is not very far from APIC result.
After few quick tests, I did not succeeded to avoid some explosions and holes of thin liquids.
I find it more simple to lower FLIP ratio.
At the end of the video, Wine example is mentioning an APIC ratio. Currently, there is none in master.
I wonder if Sebbas will add one. It does not seem to be a bad idea.
The use is explained in release notes with 2 video clips.
https://wiki.blender.org/wiki/Reference/Release_Notes/2.92/Physics
1 Like
The Mantaflow in Blender is in such a bad state barely anyone is using it. Most of parameters do not update, the results are very slow, glitchy and it crashes all the time. So I absolutely can not understand how there is enough time to implement whole new solvers but there’s no time to fix the dealbreaking bugs that cause it to be completely unusable even in the most trivial production scenarios.
AFAIK Sebastián Barschkis is the person responsible for replacing the old, but working blender’s fluid simulator with newer, but completely broken mantaflow. He’s the owner of that module, but it seems he doesn’t realize it’s not just his fun spare time project but a component of the software people use in production on daily basis.
4 Likes
+1 to @rawalanche, with every new release I give it a try hoping to find a much stable fluid simulator but there are so many random bugs and crashes that I can’t even reproduce them consistently to be able to file a bug report. 
I have been using it extensively for 1 month, for smoke and liquid, and I have not had a single crash (2.90.1, linux)
I don’t use replay mode, having to change the resolution to update is a bit of a pain …
In my tests I have not gotten crashes recently (using master).
You are reporting the problems, right?.
Even random crashes can be reported if you minimally share the “last session” recovered .blend file and the crash log report.
yea, most annoying bug, constantly changing resolution back and forth to reset the cache and update. same with changing fields settings.
Yes, I have reported many, and there are tons of them backlogged. Which really makes me confused how there can be time to implement big new features before they are solved.
In the search field of the site for reports I have searched with “liquid” or “fluid” or “mantaflow” and there are no many recent open bugs related to crashes. Could you point me some of them?
I do not doubt what you say. My intention is to test if I can reproduce the crashes on Linux to determine if the crashes you mention are related to Windows.
This is the main one so far. It’s pretty much impossible to use Mantaflow reliably as pretty much nothing that has an effect on the simulation outcome updates the replay mode correctly unless random parameter on domain is changed.
Yes, that. But you talked about crashes, and I was surprised by that because in my tests Mantaflow was quite stable with respect to not presenting crashes. I thought there might be problems with Windows version.
https://developer.blender.org/T80710 This was the one I reported last time. It’s supposedly fixed in 2.91 but that’s not a production version so I can not test it yet.