Smoke with artifacts, stripes #52768

New Issue

Dennis Fassbaender · 2017-09-15T19:43:27+02:00

2017-09-15 19:43:27 +02:00

System Information
Win7 64bit, 780GTX

Blender Version
Broken: 2.79 (and before)
Worked: (optional)

Short description of error
Smoke has artifacts/stripes!
Im not sure if this is a bug or a user fail - but ive tryed a lot of things and it seems nobody is able to help me to solve this.
So, would you please take a look why this is happening?
Its defintly NOT A RENDERBUG - its happening inside the simulation. My Domain-Material-Setup is specialy for rendering fast and make that issue better visible.
Ive added some random motion to the flowsource as well as to the turbolence-field to have some randomized motion in the smoke (thought it could fix the artifacts- but it did not help)

Exact steps for others to reproduce the error
Just play the blend file - simulate some frames and press render. The artifacting-effect will also be visible in the viewport without rendering.

DemoVideo: https://www.youtube.com/watch?v=BpCIpPENUTE
BlendFile: https://www.dropbox.com/s/6zt1rkdpvqrbu1b/Smokyfacts2.blend?dl=1

**System Information** Win7 64bit, 780GTX **Blender Version** Broken: 2.79 (and before) Worked: (optional) **Short description of error** **Smoke has artifacts/stripes!** Im not sure if this is a bug or a user fail - but ive tryed a lot of things and it seems nobody is able to help me to solve this. So, would you please take a look why this is happening? Its defintly NOT A RENDERBUG - its happening inside the simulation. My Domain-Material-Setup is specialy for rendering fast and make that issue better visible. Ive added some random motion to the flowsource as well as to the turbolence-field to have some randomized motion in the smoke (thought it could fix the artifacts- but it did not help) **Exact steps for others to reproduce the error** Just play the blend file - simulate some frames and press render. The artifacting-effect will also be visible in the viewport without rendering. DemoVideo: https://www.youtube.com/watch?v=BpCIpPENUTE BlendFile: https://www.dropbox.com/s/6zt1rkdpvqrbu1b/Smokyfacts2.blend?dl=1

Dennis Fassbaender commented

2017-09-15 19:43:27 +02:00

Changed status to: 'Open'

Dennis Fassbaender commented

2017-09-15 19:43:27 +02:00

Added subscriber: @DennisFassbaender

Dennis Fassbaender commented

2017-09-16 16:07:39 +02:00

Ive made a test with the maximal resolution of 512 and highres 2 - of course artifacts become smaller and less visible - but they are still there. It looks like smoke would have friction/is sticky to the simulation´s grid cells.
https://www.youtube.com/watch?v=EInB32aKuNo

Ive made a test with the maximal resolution of 512 and highres 2 - of course artifacts become smaller and less visible - but they are still there. It looks like smoke would have friction/is sticky to the simulation´s grid cells. https://www.youtube.com/watch?v=EInB32aKuNo

ronan ducluzeau commented

2017-09-16 20:27:09 +02:00

Added subscriber: @zeauro

ronan ducluzeau commented

2017-09-16 20:27:09 +02:00

It is a side effect of using adaptive domain.
Disable it. Recenter and scale a little bit the domain. And blockyness in simulation should be gone.

It is a side effect of using adaptive domain. Disable it. Recenter and scale a little bit the domain. And blockyness in simulation should be gone.

Dennis Fassbaender commented

2017-09-17 04:35:50 +02:00

Thank you for the quick answer.
Ive checked this with a new domain, applyed scale, disabled addaptive domain. The effect is still there.
See the screenshot:

(btw- resimulating and rendering without addaptive domain would not be a good deal as the requierd time for simulation and rendering increases very dramaticly.)

edit:
With some shader settings it is possilbe to make visible where those lines and artifacts are coming from. Its the blockyness of the smoke simulation. But why is it looking that blocky? I cant find any setting that solves that.

Thank you for the quick answer. Ive checked this with a new domain, applyed scale, disabled addaptive domain. **The effect is still there**. See the screenshot: ![Smokyfacts.jpg](https://archive.blender.org/developer/F806800/Smokyfacts.jpg) (btw- resimulating and rendering without addaptive domain would not be a good deal as the requierd time for simulation and rendering increases very dramaticly.) edit: With some shader settings it is possilbe to make visible where those lines and artifacts are coming from. Its the blockyness of the smoke simulation. But why is it looking that blocky? I cant find any setting that solves that. ![Smokyfacts2.jpg](https://archive.blender.org/developer/F807308/Smokyfacts2.jpg)

ronan ducluzeau commented

2017-09-17 12:13:20 +02:00

Stripes are not due to adaptive domain. Blockyness was due to adaptive domain .
Stripes are consequences of physics settings.

You have an emiter that moves forward and emits a smoke with an initial velocity.
Because of absence of temperature, smoke that is not slowed down by density is moving in straight line. That produces stripes.

You can increase density by using subframes for the emitter. You can inrease high resolution divisions to introduces more FFT noïse that will break more lines.
Of course, it means a simulation with better quality that takes more time.
You can modify temperature management, initial velocity of smoke.
In all cases, it means a simulation that does not look the same. It is the goal but it has consequences.

My Domain-Material-Setup is specialy for rendering fast and make that issue better visible.

Don't judge the result with same material. Use one that reduces the issue.
And don't forget to adjust volume sampling in Geometry panel of Render Tab ; it has a huge impact on rendering time.
A rigorous render would imply Motion blur because there is movement.
Fortunately, flow sampling and subframes helps to avoïd that.

With some shader settings it is possilbe to make visible where those lines and artifacts are coming from. Its the blockyness of the smoke simulation. But why is it looking that blocky? I cant find any setting that solves that.

Basic data is made of blocks, cubic cells, voxels.
The smoothness of smoke render depends of a low steps of density between neighbouring cells.
If you just globally multiply density by 25, you multiply theses steps by 25 and cells are becoming noticeable.
If you add an RGB curve after the multiply node, you can define a slope that makes them less noticeable on the contours of smoke.

The some way, if you increase thickness and slices per voxel parameter in Smoke Display settings, smoke will be blocky in Viewport.

Stripes are not due to adaptive domain. Blockyness was due to adaptive domain . Stripes are consequences of physics settings. You have an emiter that moves forward and emits a smoke with an initial velocity. Because of absence of temperature, smoke that is not slowed down by density is moving in straight line. That produces stripes. You can increase density by using subframes for the emitter. You can inrease high resolution divisions to introduces more FFT noïse that will break more lines. Of course, it means a simulation with better quality that takes more time. You can modify temperature management, initial velocity of smoke. In all cases, it means a simulation that does not look the same. It is the goal but it has consequences. > My Domain-Material-Setup is specialy for rendering fast and make that issue better visible. Don't judge the result with same material. Use one that reduces the issue. And don't forget to adjust volume sampling in Geometry panel of Render Tab ; it has a huge impact on rendering time. A rigorous render would imply Motion blur because there is movement. Fortunately, flow sampling and subframes helps to avoïd that. > With some shader settings it is possilbe to make visible where those lines and artifacts are coming from. Its the blockyness of the smoke simulation. But why is it looking that blocky? I cant find any setting that solves that. Basic data is made of blocks, cubic cells, voxels. The smoothness of smoke render depends of a low steps of density between neighbouring cells. If you just globally multiply density by 25, you multiply theses steps by 25 and cells are becoming noticeable. If you add an RGB curve after the multiply node, you can define a slope that makes them less noticeable on the contours of smoke. The some way, if you increase thickness and slices per voxel parameter in Smoke Display settings, smoke will be blocky in Viewport.

Dennis Fassbaender commented

2017-09-17 14:38:36 +02:00

Thank you for this information!

Just tryed some different things:
1- Using a higher FPS=100 + 50Subframe to have realy smooth motion = did not solve the issue
2- Tryed different setups with/without initial velocity = did not solve the issue
3- Tryed using different volume-shader setups = did not solve this issue

Maybe im doing things wrong, or blenders smokesim is not able to simulate moving inflows in productionready quality.

However- can i ask you for setting up a simple scene with a moving inflow and NO stripes/artifacts? So i can see what im doing wrong.
Just if you find time?

Best regards,
Dennis

Thank you for this information! Just tryed some different things: 1- Using a higher FPS=100 + 50Subframe to have realy smooth motion = did not solve the issue 2- Tryed different setups with/without initial velocity = did not solve the issue 3- Tryed using different volume-shader setups = did not solve this issue Maybe im doing things wrong, or blenders smokesim is not able to simulate moving inflows in productionready quality. However- can i ask you for setting up a simple scene with a moving inflow and NO stripes/artifacts? So i can see what im doing wrong. Just if you find time? Best regards, Dennis

ronan ducluzeau commented

2017-09-17 17:33:27 +02:00

SmoothedSmokyfacts2.blend

[SmoothedSmokyfacts2.blend](https://archive.blender.org/developer/F807719/SmoothedSmokyfacts2.blend)

Dennis Fassbaender commented

2017-09-17 19:06:08 +02:00

Changed status from 'Open' to: 'Resolved'

Dennis Fassbaender closed this issue

2017-09-17 19:06:08 +02:00

Dennis Fassbaender self-assigned this 2017-09-17 19:06:08 +02:00

Dennis Fassbaender commented

2017-09-17 19:06:08 +02:00