Page MenuHome
Create Task
Maniphest T79476

Mantaflow gas under 2.9 renders very different than 2.83
Needs Information from Developers, NormalPublic
Actions

Description

System Information
Operating system: Windows-10-10.0.18362-SP0 64 Bits
Graphics card: Radeon RX 570 Series ATI Technologies Inc. 4.5.13559 Core Profile Context 26.20.12028.2

Blender Version
Broken: version: 2.90.0 Alpha, branch: master, commit date: 2020-07-21 22:05, hash: rB607d745a79e0
Worked: version: 2.83.3, branch: master, commit date: 2020-07-22 06:01, hash: rB353e5bd7493e

Short description of error
The same mantaflow gas scene taken from 2.83.3 and re-simmed under 2.90.0 results in a very different sim. I can't say which is *proper*, but the differences are big enough to be problematic in trying to use it or learn to use it. It's as if the 2.83 gas is more buoyant for the same values and the 2.9 stick very close to the original particles throughout.

Exact steps for others to reproduce the error
Sim this scene under each version, saving the caches separately, since 2.90.0 caches don't work at all under 2.83.3

explosionB2.83.blend2 MBDownload

Related Objects

Event Timeline

Brian Tatosky (virtualbri) created this task.Aug 3 2020, 1:09 AM
jagsir Khokhar (jagsir) added a subscriber: jagsir Khokhar (jagsir).Aug 3 2020, 7:36 AM

blender 2.83 mantaflow sim bug,s , nobody fix manaflow problem ,please try blender 2.82a for mantaflow

Richard Antalik (ISS) changed the task status from Needs Triage to Confirmed.Aug 4 2020, 7:20 AM
Richard Antalik (ISS) added a project: Nodes & Physics.
Richard Antalik (ISS) added a subscriber: Richard Antalik (ISS).

Doesn't look like T79029 on first glance.

Sebastián Barschkis (sebbas) changed the task status from Confirmed to Needs Information from User.Aug 4 2020, 6:30 PM
Sebastián Barschkis (sebbas) added a subscriber: Sebastián Barschkis (sebbas).

I think this is in fact related to T79029. In my first answer over there I explain why fluid in 2.90 looks different compared to 2.83 and 2.82.

@Brian Tatosky (virtualbri) Can you check if that post is helpful for you?

Brian Tatosky (virtualbri) added a comment.Aug 4 2020, 7:14 PM

Hmm, I'd think if the effect of gravity was *less* in 2.9 than 2.83 my smoke would be higher and farther away from the source particles, not hanging right next to them. It's almost as if they are not getting the Initial Velocity of the source particles. I can understand the density being different; that's fine.

I *did* try some versions of emitting from a source object volume and didn't see enough differences between 2.83 and 2.9 to care, so it's now making me wonder if it's particle related.

I'll try to make an even simpler version to see if it I can determine if it's particle or initial velocity related.

Brian Tatosky (virtualbri) added a comment.Aug 4 2020, 8:07 PM


So these are sims with the buoyancy adjusted as per T79029.

This is turning off 2.83's Initial Velocity on the particles and comparing it to 2.9 with the new buoyancy settings.

I think 2.9 is ignoring the Initial Velocity on the particles enough to make drastically different looking sims.

test2.9particleemission.blend1 MBDownload
test2.8particleemission.blend1 MBDownload

REYNEP (NWERO) changed the task status from Needs Information from User to Needs Information from Developers.Aug 4 2020, 9:29 PM
REYNEP (NWERO) added a subscriber: REYNEP (NWERO).EditedAug 4 2020, 11:43 PM

Blender Version:- 2.90.0 Beta, branch: master, commit date: 2020-08-04 18:05, hash: 97be726f9319, type: Release

Brian Tatosky (virtualbri) added a comment.Aug 6 2020, 10:12 PM

Do you need anything more from me? Can I check/test anything at this time?

Brian Tatosky (virtualbri) added a comment.Aug 9 2020, 10:54 PM

Under 2.90.0 Beta, branch: master, commit date: 2020-08-07 15:11, hash: rBa93be410c974 turning the Source Initial Velocity up to 50 gets you a comparable sim to 2.83.4 sims

Luis Felipe Peredo Noguez (renderluz) added a subscriber: Luis Felipe Peredo Noguez (renderluz).Sep 24 2020, 1:09 AM

I dont know if this related, but the object velocity in 2.90.1 seems to be broken, all the other sliders on the velocity tab affect how the particles behave, but when I reduce the velocity to 0 or even -200 the particles just emit with the same speed... I may be doing something wrong, but this seems like its related to this bug report, so rather than create a new report I just thought I would confirm that object velocity seems to be off on the test I have made.