Why do Blender 2.82a, Blender 2.90.1, and Blender 2.91 smoke simulations look different? #82175

New Issue

jagsir Khokhar · 2020-10-28T06:42:08+01:00

jagsir Khokhar commented

2020-10-28 06:42:08 +01:00

System Information
Operating system: Windows-10-10.0.19041-SP0 64 Bits
Graphics card: GeForce 940MX/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.55

Short description of error
blender 2.82a, blender 2.90.1 , blender 2.91 beta mantaflow smoke simulation look different

Steps to reproduce
blender 2.82a.blend

Open file in blender 2.90 / 2.91 bake and compare results:

**System Information** Operating system: Windows-10-10.0.19041-SP0 64 Bits Graphics card: GeForce 940MX/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.55 **Short description of error** blender 2.82a, blender 2.90.1 , blender 2.91 beta mantaflow smoke simulation look different **Steps to reproduce** [blender 2.82a.blend](https://archive.blender.org/developer/F9115978/blender_2.82a.blend) Open file in blender 2.90 / 2.91 bake and compare results: ![untitled.png](https://archive.blender.org/developer/F9117410/untitled.png)

jagsir Khokhar commented

2020-10-28 06:42:08 +01:00

Added subscriber: @kkjagsir

Richard Antalik commented

2020-10-28 14:14:34 +01:00

Added subscriber: @iss

Richard Antalik commented

2020-10-28 14:14:34 +01:00

There have been intentional changes from 2.82 to 2.90 in how simulation looks. from 2.90 to 2.91 I am not aware ot any such changes.

I have checked 2.90 vs 2.91 using blender 2.82a.blend and difference is quite big:

There have been intentional changes from 2.82 to 2.90 in how simulation looks. from 2.90 to 2.91 I am not aware ot any such changes. I have checked 2.90 vs 2.91 using `blender 2.82a.blend` and difference is quite big: ![untitled.png](https://archive.blender.org/developer/F9117410/untitled.png)

Richard Antalik commented

2020-10-28 14:18:34 +01:00

Changed status from 'Needs Triage' to: 'Confirmed'

Richard Antalik commented

2020-10-28 14:19:35 +01:00

This video isn't particulary useful - it's too long, and hard to follow really, so I have removed it from report.
blender smoke.mp4

This video isn't particulary useful - it's too long, and hard to follow really, so I have removed it from report. [blender smoke.mp4](https://archive.blender.org/developer/F9115965/blender_smoke.mp4)

jagsir Khokhar commented

2020-10-29 07:59:15 +01:00

I am waiting for a long time to fix blender mantaflow this bug,
now tell me when blender team fixes this bug?

I am waiting for a long time to fix blender mantaflow this bug, now tell me when blender team fixes this bug?

Tom Wheeler commented

2020-10-29 18:06:20 +01:00

Added subscriber: @tomwheeler

Jacques Lucke commented

2020-11-03 12:54:37 +01:00

Added subscribers: @sebbas, @JacquesLucke

Jacques Lucke commented

2020-11-03 12:54:37 +01:00

This is caused by 663e047102.
It seems to be an intended side effect of the change, because the commit message does not mention that this could happen.
Therefore I'm setting this as high priority, because it should be fixed or document before the release.

This is caused by 663e047102. It seems to be an intended side effect of the change, because the commit message does not mention that this could happen. Therefore I'm setting this as high priority, because it should be fixed or document before the release.

Sebastián Barschkis was assigned by Dalai Felinto

2020-11-03 18:27:09 +01:00

Sebastián Barschkis changed title from ~~why blender 2.82a, blender 2.90.1 , blender 2.91 beta mantaflow smoke simulation look different ?~~ to Why do Blender 2.82a, Blender 2.90.1, and Blender 2.91 smoke simulations look different?

2020-11-05 19:01:02 +01:00

Sebastián Barschkis commented

2020-11-05 19:27:27 +01:00

Changed status from 'Confirmed' to: 'Resolved'

Sebastián Barschkis closed this issue

2020-11-05 19:27:27 +01:00

Sebastián Barschkis commented

2020-11-05 19:27:27 +01:00

Some explanations:

2.82a: looks different for 2 reasons:
- Volumes are rendered slightly different starting from 2.83 (D6951)
- Buoyancy / gravity is treated differently. For similar results, 2.9x simulations need more buoyancy than 2.8x simulations.
2.90.1: has a bug with adaptive domains which makes smoke rise too quickly (compared to the same non-adaptive domain). The behavior of adaptive domains in 2.90.1 is incorrect.
2.91: should resolve the issues. There was an attempt to fix the issues from 2.90.1 (663e047102). This was not enough however. There is now an additional change (60859d8912) in 2.91 to complete the fix.

Some explanations: - **2.82a:** looks different for 2 reasons: - Volumes are rendered slightly different starting from 2.83 ([D6951](https://archive.blender.org/developer/D6951)) - Buoyancy / gravity is treated differently. For similar results, 2.9x simulations need **more** buoyancy than 2.8x simulations. - **2.90.1:** has a bug with adaptive domains which makes smoke rise too quickly (compared to the same non-adaptive domain). The behavior of adaptive domains in 2.90.1 is **incorrect**.![290_wrong.png](https://archive.blender.org/developer/F9211292/290_wrong.png) - **2.91:** should resolve the issues. There was an attempt to fix the issues from 2.90.1 (663e047102). This was not enough however. There is now an additional change (60859d8912) in 2.91 to complete the fix.![291_correct.png](https://archive.blender.org/developer/F9211295/291_correct.png)

Sign in to join this conversation.

No Label

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Why do Blender 2.82a, Blender 2.90.1, and Blender 2.91 smoke simulations look different? #82175