Page MenuHome

Mantaflow -> random missing square-shaped chunks in generated foam particle cache
Confirmed, NormalPublicBUG

Description

System Information
Operating system: Windows 10 Pro x64
CPU: Intel i7 8700k
RAM: 64GB DDR4-3200
Graphics card: Geforce GTX1070

Blender Version
Broken: 2.91.2 (Steam) & 2.92 Beta Daily Builds from 06. - 11. Feb. 2021
(what i have tested so far)

Short description of error
Foam generated by Mantaflow has random square chunks missing. These missing chunks randomly change position (mostly per frame, but some stay in position longer ... ~5 Frames). Denser foam seems to generate visibly more of these.

Found two similar bugreports, but not directly related to this specific issue ... maybe a grid-issue too?

Exact steps for others to reproduce the error
Its a simple scene (flood gate simulation), fluid domain, two fluid emitters (for the different water heights), some enviromental assets (fluid affectors -> collision ... a wall, a floor, four gates and a boat ... not related to the issue besides generating foam via fluid collision) ... short overview ...

Scene:

  • 3840x2160px
  • 50FPS

Fluid Domain:

  • Size = 3.47m x 19.9m x 1.81m
  • Res. = 256
  • timesteps minimum = 2
  • delete in obstacle = enabled
  • generate foam particles only (no splash, no bubbles)
  • everything else uses default values

Cache:

  • manually set destination (external USB3 HDD)
  • Frames = 1 - 2000;
  • Modular (resumable)
  • Format Volumes = OpenVDB
  • final baked cache (guide particles + foam particles + mesh) = ~160 GB

First time i noticed this issue after baking foam and mesh (Blender 2.91.2)

A few days later (today) after further working on the scene and having switched to 2.92 Daily Builds (hoping anything would change)


Top Down View (foam particles only)

I would have provided videofootage too (showing how the holes change based on current frame), but nvidia shadowplay refuses to record blender for some reason (record desktop is enabled)

Here is the original project file (before i switched to 2.92 beta), but be warned, baking this takes some time...

Kind Regards

Event Timeline

Sebastián Barschkis (sebbas) changed the task status from Needs Triage to Confirmed.Thu, Feb 11, 10:51 AM

Hi, thanks for the report! From the top of my head this looks like the problem from T79995.

Definitely a big issue, I will make this a priority for 2.93.

Hey Sebastián, thanks for replying so quickly and the great work you do.
I'm not a dev, but i don't think this issue is directly related/comparable to T79995.

The problem is not nessesary the grid beeing visible due to low particle upres settings, it's more that the grid is missing data at certain coordiantes (data-fragmentation/-corruption?!?), which also varies from frame to frame ... the foam in my example-images also contains much thinner foam-density-areas (where i would expect this problem to happen more frequent) without any issue at all. It's for some reason more concentrated/frequent in thicker density areas within the foam :/

But again, i'm not a dev, could be anything ...

Bastien Montagne (mont29) changed the subtype of this task from "Report" to "Bug".