Page MenuHome

Fire physics break when high resolution smoke activated (2.8)
Open, Confirmed, LowPublic

Description

System Information
Operating system: Windows-10-10.0.17134 64 Bits
Graphics card: GeForce GTX 970/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 419.67
CPU: Intel i5 4590k

Blender Version
Broken: version: 2.80 (sub 57), branch: master, commit date: 2019-04-16 20:13, hash: rB5d8030b8f3cf

Short description of error
The physics involved with fire and smoke simulations drastically change when 'High Resolution' is ticked in the domain object's settings. Specifically, with the collision object that I'm using as a ground plane for the flames to make contact with, and run up against. The flames look fine and appear to be doing what I want, right up until it tries to go High Res for more fidelity. Afterwards, the flames cling to the floor - even appearing to sink like dry ice rather than come out as a stream as they had done before. See screenshots here:

(Before high res - looks right)

(Same frame but with high res checked. Something amiss.)

Exact steps for others to reproduce the error

  1. After opening .blend file attached below, allow physics to bake (either by clicking bake or by allowing animation to play) until about frame ~60 using current settings. Should look fine.
  2. Check box marked 'High Resolution'. Ensure that subdivisions are set to 1 (this number isn't important, but keep it low just for the sake of speed) and bake again.
  3. Results should be noticeably different around frame 30 or even earlier. It's as though the maximum force field bounds are not being respected, despite the fact that the high resolutions setting shouldn't really (I would have thought) interfere with the physics bake - as the flames now cower away from Force force before entering its sphere of influence. Either that or the flame's burning temperature is altered, to stop flames from rising (in this case, they drop rather than rise).

If this is not a bug, I apologise.

Details

Type
Bug

Event Timeline

Sebastian Parborg (zeddb) triaged this task as Confirmed, Low priority.

This happens with 2.79 too. The smoke sim is to be replaced with mantaflow soonish, so this is not a high priority.

Are developers considering merging mantaflow into blender for 2.80 release? If not, I think these bugs should be addressed. As it is, we have a broken simulation system that should not be shipping with an official release.