Page MenuHome

Mantaflow fire disappears when second smoke-only emitter enabled
Closed, ResolvedPublic

Description

System Information
Operating system: Windows-10-10.0.18362-SP0 64 Bits
Graphics card: GeForce GTX 1070 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 431.86

Blender Version
Broken: version: 2.82 (sub 5), branch: fluid-mantaflow, commit date: 2019-12-16 14:05, hash: rBed32db06c124
Worked: (optional)

Short description of error
[Smoke & Fire emitter seems to only emit smoke after a second emitter (smoke only) is added.
Here's the same Sim without & with the second emitter enabled-
]

Exact steps for others to reproduce the error
[Bake the attached without the 'ParticleEmit' object enabled and you should see fire from the 'Emitter' object, bake again with the 'ParticleEmit' object enabled and the fire disappears.

]
[Based on the default startup or an attached .blend file (as simple as possible)]

Event Timeline

One question, how do you make the second smoke flow simulated?
I followed the described steps of enabling Particle Emit and bake again, but that made no difference.

Mark Spink (marks) added a comment.EditedTue, Jan 14, 12:41 AM

You've got me! The video above is (rendered) exactly the result of the simulation in the .blend file I uploaded, without and with the second emitter enabled.

Are you looking at the result in Solid or Rendered viewport mode?
As in, Blender smoke looks shit/tiny unless you have high density in the shader so the smoke is barely visible in solid mode once simulated.

Or is this another Linux v Windows thing? Maybe the Developer is using Linux as well, and the Windows build is more broken than Linux?

Have you re-baked the particle system on the second emitter? Particles are (still!) very, very, very rubbish in Blender, the only way I can render across multiple machines vaguely reliably is to bake, without 'disk cache' or 'external' enabled. If the particles aren't there there is no second emitter.

Here's a few grabs of my interface & some screen capture showing the bake without then with the second enitter enabled-



As you can see in the stills & the video: zero flames when the second emitter is in play...

What country are you in? (I'm not after a date!) I'd be happy to Facetime / Whatsapp / whatever video call you to show you what I'm getting in real-time, although the capture video sort-of says it all...
(and I am in Sydney so a pain in the arse timezone wise)

I use windows too. It's weird, but I can't do the second simulation.

Is it really the same file?
Can you try the latest daily build: https://builder.blender.org/download/

Curios-er and curios-er!!!
I'm getting closer to giving up on Mantaflow for smoke all together!...

The build I reported this bug from: commit date: 2019-12-16 14:05, hash: rBed32db06c124 is the current build AFAIK (I keep looking for a new one, with hopefully some fixes, but the blender.org page still shows 16th December build as the download).
I only use builds downloaded from blender.org since they became available there: trying to keep some consistency, and less multiple installations of Blender!

I have just re-built the scene in the November 26 build (what I'm calling the 'working' build render wise- see T72730 an obviously inconsistent bug, as I can render fire in the 16 December build in this bug, until I enable the second emitter)-
and it works OK (sort-of)-

Here's the 26 November build .blend-


Don't try it in the current 16 December build as it won't 'see' any of the fluid modifiers from the older builds, and you'll have to re-add them and re-create all the key-frames on them!

Do you get different results from me if you try the .blend I sent in the T72730 bug? Are you using the same version of Windows with Nvidia GPU/GPUs? Are you using an older Nvidia driver as the newest one ruins Blackmagic RAW decoding in Davinci Resolve?
(Do you live in the southern hemisphere? Have you got blue eyes? I'm clutching at straws here!)

I just tried the 16 December build file (first .blend uploaded to this report) on 2 other machines: an HP Z800 with 2 Xeons and a single GTX1070 & Windows 10 pro, and a Gigabyte Aero15 laptop with an i7 and a GTX 1060 and Windows 10 Home- and got the same results on both as I do on my Threadriper (as used in all the above, and shown in the 'System Information' bit at the top of this report).

Fuck! I'm a broadcast engineer by day, and if anything I used at work was this inconsistent I'd throw the bastard in a skip and set fire to it!!!

I don't know what to say, you can clearly see what happens when I try this as I recorded my screen whilst doing it! And that was definitely using the 12 December build.

Sorry!..........

Mark Spink (marks) added a comment.EditedWed, Jan 15, 6:16 AM

Interestingly... I just downloaded 2.83 Alpha, and it behaves like you are seeing, nothing happens when the particle emitter is enabled (apart from crashing the first time I baked, but not the second: WTF?!), I assume this means that the Mantaflow branch download is now effectively out of date (too bloody confusing! do I use 2.82 beta, 2.83 alpha or Mantaflow experimental to do/test smoke?!?).
I notice that the particle system is ghosted when the scene is opened, and show no 'cache' in the particle panel.
I shall try re-making the particle system afresh and see what happens...

Yep: it's the extremely rubbish particle systems! Add a new particle system and copy all the parameters from the original (ghosted) one, then disable / delete the old one, and chose the new particle settings in the physics/fluid tab and it will work again.

And in 2.83 Alpha it DOES render fire (as does 2.82 beta- just downloaded & tried that too).

Which begs the question- WHY THE FUCK is Mantaflow still up as an experimental download if it is-
1/out of date, and
2/not experimental any more?

Please God remove it from the downloads page - I'd rather report REAL bugs and not waste peoples time!
I'm off to re-test ALL my extant Mantaflow bugs in the actually current version, which I'm assuming is 2.82 beta?
Or will the version in 2.83 Alpha now be the most up to date version of Mantaflow, as in the one to test past / present bugs with????? (HELP!)...

Please close this bug as I think it's now working OK from what I can see.
Although I think there is still an issue with the fire in Mantaflow, as in it used to fade out when the fuel was spent, and now it seems to just go like a light being turned off, but I'll save that for another bug report...

Sorry to waste your time!!!