Overlapping volumes with variable density show artifacts #115871

New Issue

Sebastien Larocque · 2023-12-07T05:08:59+01:00

2023-12-07 05:08:59 +01:00

System Information
Operating system: Windows-10-10.0.19045-SP0 64 Bits
Graphics card: NVIDIA GeForce RTX 3080 Ti/PCIe/SSE2 NVIDIA Corporation 4.6.0 NVIDIA 546.01

Blender Version
Broken: version: 4.0.2, branch: blender-v4.0-release, commit date: 2023-12-05 07:41, hash: 9be62e85b727
Worked: Never

Short description of error
Overlapping volumes with variable density show artifacts

Exact steps for others to reproduce the error

Open the attached file here.
Follow the instruction below to reproduce the different scenarios. Each scenario is represented with a different atmosphere object (and shader) where there are overlapping cloud objects with different configurations. This results in several tests for each atmosphere.

This bug report is related to a global problem that I shared several months ago with overlapping volumetric objects. The problem is essentially the same, but I misunderstood the nature of the problem at that time.
Reference: #107418.

I decided to create a new bug report because I have been able to isolate a precise aspect causing the problem. Nonetheless, the other bug report is still valid because I described several problematic aspects of the overlapping volumetric objects.

The first bug report highlighted the problem without providing good explanation and solutions. However, one answer provided explanations and a partial solution to one case or issue, but this new bug report is able to demonstrate exactly how to reproduce certain artifacts. Note that the problem might be more complex than what is described here.

What causes the artifacts?

I noticed that the artifacts are present in the smaller volumes (clouds) when the biggest volume (atmosphere) has different volume density values in it at different places. Sometimes, it is darker, sometimes brighter. Here are two ways to reproduce the problem.

Variable height fog/haze with the atmosphere volume. The volume density value is adjusted based on the height position.
Creating a noise pattern in the atmosphere volume. Using a constant volume density value multiplied by a binary noise 3D texture to make the fog/haze cleared. This is necessary for an atmosphere because the minimal volume density value is already too high to compute the right fog/haze value in the distance and to make it clearer with the altitude.

Note: It works correctly with volumetric clouds (variable density) and a volumetric atmosphere where the density is constant. However, this is not usable because a real atmosphere has exponential density and Blender smallest density volumetric value is already too high. We need to multiply by a noise pattern to make it clearer.

Objects in the scene

Here’s a description of the objects and their role in this test.

“Common” collection:
It holds the objects common to all tests, mainly the clouds objects where the artifacts are visible. They are overlapping in different ways with the different atmosphere test objects described below. In this collection:

Cloud_1_AboveAndUnderAtmosphere: A cloud extending above and under the atmosphere volume. Used to test with the limits of the atmosphere.

Cloud_1_InsideAtmosphere: A cloud completely contained within the atmosphere. Typical scenario for building scenes.

Cloud_EmptyTest: A volume with zero density with the same size as “Cloud_1_InsideAtmosphere”.

Cloud_EmptyTest_Lower: Same as “Cloud_EmptyTest”, but the position is lower. This object allows us to see the artifact better because the top face is lower in the atmosphere. There’s more density and it might be the reason why it’s more visible.

Cloud_EmptyTest_VariableSurfaceHeight: This object is similar to “Cloud_EmptyTest”. It does the same thing, but the top face makes a slope to overlap with the atmosphere at different altitudes. It illustrates the same point as “Cloud_EmptyTest_Lower”, but with variable heights.

Cloud_Sphere: A test cloud (very basic shape) in a cube where the upper part is above the atmosphere.

“Atmosphere - Tests” collection:
It holds the different scenarios of this test project. You should enable only one atmosphere object at a time. Each object generally tests one aspect related to the cloud objects found in the “Common” collection. Each atmosphere object overlaps with all the clouds at once testing many scenarios.

Atmosphere_ConstantDensity: This works correctly. The density is constant in every position of the atmosphere. Actually, this is the only way it works without artifacts with the overlapping clouds.

Atmosphere_ConstantDensityWithVariableAbsortion: Same as “Atmosphere_ProgressiveDensity”, but the volume scatter density is constant. The aspect that is variable is the density of the “Volume Absorption” node. This illustrates once again the artifact problem with the variable density, but in another volume node.

Atmosphere_ProgressiveDensity: This is where the main point of this bug report is demonstrated. The atmosphere has a density that varies based on the height. I believe this is where the focus should be to investigate the source of the problem.

Atmosphere_ProgressiveDensity_Narrow: Same thing as “ ProgressiveDensity”, but the width is smaller and let pass more sunlight.

Atmosphere_WorkaroundWithNoise: I worked to make the atmosphere density clearer with the noise texture strategy. It worked for that, but it introduced the artifacts, probably because of the variable density at different positions.

Note: If you enable this atmosphere and change the scale of the “Noise Texture” node from 1.0 to 0.002, the noise is very large and you can see the artifacts differently because of the bigger voids between the density parts.

“Atmosphere 2 - Tests” collection:
It can be enabled instead of the “Atmosphere - Tests”. This tests two atmospheres side by side and overlapping the “Cloud_EmptyTest” and “Cloud_EmptyTest_Lower” objects. They are precisely a working and none-working atmospheres. Nothing particular happened here, but I wanted to test the overlapping clouds with a working and none-working objects to see how they would change their appearance. Everything is displayed as expected, but this could be used for more advanced test.

Observations

Observations 1:
The distance between the top face of the cloud objects and the top face of the atmosphere has an influence. You can look at the scene setup for more details. In a nutshell, I created “Cloud_EmptyTest” where the top face is barely visible and I decided to place aside the same object, but a little bit lower. The top face seems to have more density. I also created “Cloud_EmptyTest_VariableSurfaceHeight” to illustrate the density visibility of the artifacts where the height of the top face is variable. This scenario can be reproduced by enabling the atmosphere “Atmosphere_ProgressiveDensity”.

Observations 2:
Related to the screenshot “ViewportView_PerspectiveLeft_OrthogonalRight.jpg”
This image shows as a collage side by side two views taken from the viewport. The left view is the normal perspective view and the right view is set to orthographic. While in orthographic view, the empty cloud volume “Cloud_EmptyTest” that has a surface near the top of the atmosphere volume is barely visible (need to crank up the contrast to see it). While in perspective mode, we can see the artifacts on the four sides of the empty cloud cube, which is not possible with the top-down orthographic camera here.

It appears to be related to the direction of the variable density in the atmosphere. If you tilt the camera while in orthographic mode, the sides of the cube show the artifacts.

This can be reproduced by enabling the atmosphere “Atmosphere_ProgressiveDensity” or “Atmosphere_WorkaroundWithNoise”.

Observations 3:
The objects “Cloud_EmptyTest” and “Cloud_1_InsideAtmosphere” have the same size and are at the same height in the atmosphere. For an unknown reason, “Cloud_1_InsideAtmosphere” shows darker artifacts in the empty area around the cloud while “ Cloud_EmptyTest” shows darker artifacts on the side faces while having a very subtle darker shade in the central area.

Very important note:
I noted in the past (including my first bug report mentioned above) that the artifacts seems to appear on the surfaces and not exactly within the volume as original thought.

Observations 4:
“Cloud_Sphere” object. I created a cloud in the shape of a sphere. The top part of it is above (outside) the atmosphere. It looks like there is some artifact at the edge of the atmosphere top face, but I believe the atmosphere makes the light more diffused around the sphere, which creates a cutting shadow at the edge of the atmosphere. This is a test and the appearance is probably normal.

Observations 5:
Shadows. The objects with the artifacts do not project more shadows. The object “Cloud_Sphere” projects a clear shadow because of the density and the other clouds also project some shadow. Nowhere, there is a shadow visible because of what appears the extra density at the surface of the cloud objects. As I mentioned, there is a darker or brighter shade that appears more at the surface. It does not seem to create extra density either that could create shadow. It does not block the light. The artifacts are visible to the camera, but don’t seem to influence the luminosity of the atmosphere whether they are present or not.

Furthermore, in this test project, the artifacts are visible from the top-down view and if you adjust the camera to rotate downwards until you get the horizontal view with the sun visible, you can see the artifacts disappearing. That makes me believe that the camera angle has something to do with it.

Observations 6:
Render pass. In the viewport, there’s the option “Viewport Shading”. Under it, you’ll find the “Render Pass” drop-down menu. I suggest looking at every render pass when using the atmosphere “Atmosphere_ProgressiveDensity”. Some will show the artifacts. Maybe it’s possible to see in which render passes the problem is to provide more clues.

My analysis and some hypothesis

I cannot exactly explain the problem behind, but in surface it appears clear to me if I have two overlapping volumes and both have variable density, I’m going to have artifacts.

I have this impression that there is a density value obtained from the atmosphere for the cloud objects when overlapping. This value is used in every position of the cloud objects without being reevaluated with the changing atmosphere density based on the height. However, I have some doubt because as I explained above, the visibility of the artifacts changes with the camera angle.

Conclusion

It’s important to keep in mind that I highlighted several variables in my first bug report and also the difficulty to reproduce the problem. Very often, the problem is mostly visible from a certain camera angle, sun light angle within a certain density range in the atmosphere and under certain step rates settings.

This new bug report is not meant to replace the first one, but to focus on one particular aspect, density variability, where the problem can be reproduced and to provide a solid approach to solving it in the conditions described above.

Furthermore, I’m still testing different scenarios because it’s not obvious why the problem is more or less apparent depending on the camera angle with the affected volumes.

I understand this is a problem that seems quite big and has been present for years without having been solved according to many other bug reports related. I hope this new bug report will provide key information to fix it once for all. Do you think you can investigate the source of the problem more precisely with this new bug report?

**System Information** Operating system: Windows-10-10.0.19045-SP0 64 Bits Graphics card: NVIDIA GeForce RTX 3080 Ti/PCIe/SSE2 NVIDIA Corporation 4.6.0 NVIDIA 546.01 **Blender Version** Broken: version: 4.0.2, branch: blender-v4.0-release, commit date: 2023-12-05 07:41, hash: `9be62e85b727` Worked: Never **Short description of error** Overlapping volumes with variable density show artifacts **Exact steps for others to reproduce the error** * Open the attached file here. * Follow the instruction below to reproduce the different scenarios. Each scenario is represented with a different atmosphere object (and shader) where there are overlapping cloud objects with different configurations. This results in several tests for each atmosphere. This bug report is related to a global problem that I shared several months ago with overlapping volumetric objects. The problem is essentially the same, but I misunderstood the nature of the problem at that time. Reference: https://projects.blender.org/blender/blender/issues/107418. I decided to create a new bug report because I have been able to isolate a precise aspect causing the problem. Nonetheless, the other bug report is still valid because I described several problematic aspects of the overlapping volumetric objects. The first bug report highlighted the problem without providing good explanation and solutions. However, one answer provided explanations and a partial solution to one case or issue, but this new bug report is able to demonstrate exactly how to reproduce certain artifacts. Note that the problem might be more complex than what is described here. **What causes the artifacts?** I noticed that the artifacts are present in the smaller volumes (clouds) when the biggest volume (atmosphere) has different volume density values in it at different places. Sometimes, it is darker, sometimes brighter. Here are two ways to reproduce the problem. 1. Variable height fog/haze with the atmosphere volume. The volume density value is adjusted based on the height position. 2. Creating a noise pattern in the atmosphere volume. Using a constant volume density value multiplied by a binary noise 3D texture to make the fog/haze cleared. This is necessary for an atmosphere because the minimal volume density value is already too high to compute the right fog/haze value in the distance and to make it clearer with the altitude. Note: It works correctly with volumetric clouds (variable density) and a volumetric atmosphere where the density is constant. However, this is not usable because a real atmosphere has exponential density and Blender smallest density volumetric value is already too high. We need to multiply by a noise pattern to make it clearer. **Objects in the scene** Here’s a description of the objects and their role in this test. “Common” collection: It holds the objects common to all tests, mainly the clouds objects where the artifacts are visible. They are overlapping in different ways with the different atmosphere test objects described below. In this collection: Cloud_1_AboveAndUnderAtmosphere: A cloud extending above and under the atmosphere volume. Used to test with the limits of the atmosphere. Cloud_1_InsideAtmosphere: A cloud completely contained within the atmosphere. Typical scenario for building scenes. Cloud_EmptyTest: A volume with zero density with the same size as “Cloud_1_InsideAtmosphere”. Cloud_EmptyTest_Lower: Same as “Cloud_EmptyTest”, but the position is lower. This object allows us to see the artifact better because the top face is lower in the atmosphere. There’s more density and it might be the reason why it’s more visible. Cloud_EmptyTest_VariableSurfaceHeight: This object is similar to “Cloud_EmptyTest”. It does the same thing, but the top face makes a slope to overlap with the atmosphere at different altitudes. It illustrates the same point as “Cloud_EmptyTest_Lower”, but with variable heights. Cloud_Sphere: A test cloud (very basic shape) in a cube where the upper part is above the atmosphere. “Atmosphere - Tests” collection: It holds the different scenarios of this test project. You should enable only one atmosphere object at a time. Each object generally tests one aspect related to the cloud objects found in the “Common” collection. Each atmosphere object overlaps with all the clouds at once testing many scenarios. Atmosphere_ConstantDensity: This works correctly. The density is constant in every position of the atmosphere. Actually, this is the only way it works without artifacts with the overlapping clouds. Atmosphere_ConstantDensityWithVariableAbsortion: Same as “Atmosphere_ProgressiveDensity”, but the volume scatter density is constant. The aspect that is variable is the density of the “Volume Absorption” node. This illustrates once again the artifact problem with the variable density, but in another volume node. Atmosphere_ProgressiveDensity: This is where the main point of this bug report is demonstrated. The atmosphere has a density that varies based on the height. I believe this is where the focus should be to investigate the source of the problem. Atmosphere_ProgressiveDensity_Narrow: Same thing as “ ProgressiveDensity”, but the width is smaller and let pass more sunlight. Atmosphere_WorkaroundWithNoise: I worked to make the atmosphere density clearer with the noise texture strategy. It worked for that, but it introduced the artifacts, probably because of the variable density at different positions. Note: If you enable this atmosphere and change the scale of the “Noise Texture” node from 1.0 to 0.002, the noise is very large and you can see the artifacts differently because of the bigger voids between the density parts. “Atmosphere 2 - Tests” collection: It can be enabled instead of the “Atmosphere - Tests”. This tests two atmospheres side by side and overlapping the “Cloud_EmptyTest” and “Cloud_EmptyTest_Lower” objects. They are precisely a working and none-working atmospheres. Nothing particular happened here, but I wanted to test the overlapping clouds with a working and none-working objects to see how they would change their appearance. Everything is displayed as expected, but this could be used for more advanced test. **Observations** Observations 1: The distance between the top face of the cloud objects and the top face of the atmosphere has an influence. You can look at the scene setup for more details. In a nutshell, I created “Cloud_EmptyTest” where the top face is barely visible and I decided to place aside the same object, but a little bit lower. The top face seems to have more density. I also created “Cloud_EmptyTest_VariableSurfaceHeight” to illustrate the density visibility of the artifacts where the height of the top face is variable. This scenario can be reproduced by enabling the atmosphere “Atmosphere_ProgressiveDensity”. Observations 2: Related to the screenshot “ViewportView_PerspectiveLeft_OrthogonalRight.jpg” This image shows as a collage side by side two views taken from the viewport. The left view is the normal perspective view and the right view is set to orthographic. While in orthographic view, the empty cloud volume “Cloud_EmptyTest” that has a surface near the top of the atmosphere volume is barely visible (need to crank up the contrast to see it). While in perspective mode, we can see the artifacts on the four sides of the empty cloud cube, which is not possible with the top-down orthographic camera here. It appears to be related to the direction of the variable density in the atmosphere. If you tilt the camera while in orthographic mode, the sides of the cube show the artifacts. This can be reproduced by enabling the atmosphere “Atmosphere_ProgressiveDensity” or “Atmosphere_WorkaroundWithNoise”. Observations 3: The objects “Cloud_EmptyTest” and “Cloud_1_InsideAtmosphere” have the same size and are at the same height in the atmosphere. For an unknown reason, “Cloud_1_InsideAtmosphere” shows darker artifacts in the empty area around the cloud while “ Cloud_EmptyTest” shows darker artifacts on the side faces while having a very subtle darker shade in the central area. Very important note: I noted in the past (including my first bug report mentioned above) that the artifacts seems to appear on the surfaces and not exactly within the volume as original thought. Observations 4: “Cloud_Sphere” object. I created a cloud in the shape of a sphere. The top part of it is above (outside) the atmosphere. It looks like there is some artifact at the edge of the atmosphere top face, but I believe the atmosphere makes the light more diffused around the sphere, which creates a cutting shadow at the edge of the atmosphere. This is a test and the appearance is probably normal. Observations 5: Shadows. The objects with the artifacts do not project more shadows. The object “Cloud_Sphere” projects a clear shadow because of the density and the other clouds also project some shadow. Nowhere, there is a shadow visible because of what appears the extra density at the surface of the cloud objects. As I mentioned, there is a darker or brighter shade that appears more at the surface. It does not seem to create extra density either that could create shadow. It does not block the light. The artifacts are visible to the camera, but don’t seem to influence the luminosity of the atmosphere whether they are present or not. Furthermore, in this test project, the artifacts are visible from the top-down view and if you adjust the camera to rotate downwards until you get the horizontal view with the sun visible, you can see the artifacts disappearing. That makes me believe that the camera angle has something to do with it. Observations 6: Render pass. In the viewport, there’s the option “Viewport Shading”. Under it, you’ll find the “Render Pass” drop-down menu. I suggest looking at every render pass when using the atmosphere “Atmosphere_ProgressiveDensity”. Some will show the artifacts. Maybe it’s possible to see in which render passes the problem is to provide more clues. **My analysis and some hypothesis** I cannot exactly explain the problem behind, but in surface it appears clear to me if I have two overlapping volumes and both have variable density, I’m going to have artifacts. I have this impression that there is a density value obtained from the atmosphere for the cloud objects when overlapping. This value is used in every position of the cloud objects without being reevaluated with the changing atmosphere density based on the height. However, I have some doubt because as I explained above, the visibility of the artifacts changes with the camera angle. **Conclusion** It’s important to keep in mind that I highlighted several variables in my first bug report and also the difficulty to reproduce the problem. Very often, the problem is mostly visible from a certain camera angle, sun light angle within a certain density range in the atmosphere and under certain step rates settings. This new bug report is not meant to replace the first one, but to focus on one particular aspect, density variability, where the problem can be reproduced and to provide a solid approach to solving it in the conditions described above. Furthermore, I’m still testing different scenarios because it’s not obvious why the problem is more or less apparent depending on the camera angle with the affected volumes. I understand this is a problem that seems quite big and has been present for years without having been solved according to many other bug reports related. I hope this new bug report will provide key information to fix it once for all. Do you think you can investigate the source of the problem more precisely with this new bug report?

Overlapping volumes with variable density.blend

1.8 MiB

Atmosphere_ConstantDensity.jpg

547 KiB

ViewportView_PerspectiveLeft_OrthogonalRight.jpg

128 KiB

Atmosphere_ProgressiveDensity.jpg

555 KiB

Sebastien Larocque added the

labels 2023-12-07 05:08:59 +01:00

Sebastien Larocque referenced this issue

2023-12-07 05:10:14 +01:00

Visible density seen inside an overlapping volume with zero density value (Cycles) #107418

Pratik Borhade commented

2023-12-07 07:15:30 +01:00

Hi, thanks for the report. I can confirm the described problem about artifacts with variable density. We already have decent number of reports about Overlapping volumes but we would need to check them if it's causing due to variable density or not.
I would classify Atmosphere_WorkaroundWithNoise and Atmosphere_ProgressiveDensity as same since both has different density at every position.

@LukasStockner , I think you were investigating some bugs report about overlapping volume. Can you check this as well?
Quick way to observe this issues is: only enable Atmosphere_ProgressiveDensity from atmosphere collection then hit render.

Hi, thanks for the report. I can confirm the described problem about artifacts with variable density. We already have decent number of reports about Overlapping volumes but we would need to check them if it's causing due to variable density or not. I would classify `Atmosphere_WorkaroundWithNoise` and `Atmosphere_ProgressiveDensity` as same since both has different density at every position. @LukasStockner , I think you were investigating some bugs report about overlapping volume. Can you check this as well? Quick way to observe this issues is: only enable `Atmosphere_ProgressiveDensity` from atmosphere collection then hit render.

Pratik Borhade added

and removed

labels 2023-12-07 07:15:52 +01:00

Sebastien Larocque commented

2024-04-06 23:08:37 +02:00

This is an update with new tests related to this bug report. This update brings new elements and probably a new point of view to debug the issue.

I found this when modifying the “Max Steps” where the camera was within certain angles with the volumes. I included some screenshots here and a new test project to test 3 precise cases.

Observations:

The parameter “Render | Volumes | Max Steps” has an influence.

Moving “Camera_Perspective” on the Y axis. The artifacts appears from a certain distance. Does the distance have something to do with the artifacts?

I created other camera “Camera_Perspective2” that is more distant with some zoom in. Instead of Max Steps at 24, the difference in the problem is mainly visible at Max Steps 28. There is something with the distance.

I created an orthographic camera. The artifact problem between the cubes seems to be present at certain angles, but not the contrast in lower or higher density based on the camera distance from the volumes.

Tests:

There are 3 tests for this.

Set the viewport shading to “Rendered” to have the Cycles view.
Enable camera “Camera_Perspective”.

Test 1.

Enable “Atmosphere_GN_Test_ProblemAt4MaxSteps”
Set Max Steps to 4.
Because the volume is not divided into many adjacent cubes, there are no artifacts visible between the cubes. However, at the bottom of the image (or closest part of the volume), we can see a higher density. It seems still related to the distance because moving the camera changes the transition area between the lower and higher density.

Test 2.

Enable “Atmosphere_GN_Test_ProblemAt24MaxSteps” instead of “Atmosphere_GN_Test_ProblemAt4MaxSteps”
Set Max Steps to 24.
The same problem but at a higher Max Steps and with smaller adjacent volumes.

Note: Changing the Max Steps to the default of 1024 and changing the Step Rate to the default value of 1.00 makes the artifacts visible again between the cubes. Changing the Step Rate to 0.1 makes the artifact disappear.

Sadly, changing to 0.1 does not solve the problem for all scenarios. In my original project, I used this Step Rate and I had artifacts between the cubes for a cloud layer in high altitude. I also noticed that the problem was visible for density values like 0.02. To avoid the problem, I was limited to values a lot less dense like 0.0002. This might also explain why including VDB clouds in a volume with variable density also creates artifacts around the edges.

Test 3.

Enable “Camera_Orthographic”.
Enable “Atmosphere_GN_Test_ProblemAt24MaxSteps”.
Rotate the camera using the parent object “Rotate” for the orthographic camera using the X axis. At 8 degrees, the artifacts start to become visible. Increasing the rotation brings the camera lower to the ground.
In the case of an orthographic camera, the angle is a factor to make the artifacts visible or not.

Conclusion

That seems to only happen for volumes of variable density. The initial project in this bug report highlighted the problem for overlapping volumes with variable density. This update highlights the aspect of the Max Steps with non-overlapping volumes.

Furthermore, I reported this problem in another bug report under similar circumstances. Since then, it has been around a year and no follow-up has been done. Is this problem being investigated by the Blender team?

This is an update with new tests related to this bug report. This update brings new elements and probably a new point of view to debug the issue. I found this when modifying the “Max Steps” where the camera was within certain angles with the volumes. I included some screenshots here and a new test project to test 3 precise cases. **Observations:** 1. The parameter “Render | Volumes | Max Steps” has an influence. 2. Moving “Camera_Perspective” on the Y axis. The artifacts appears from a certain distance. Does the distance have something to do with the artifacts? 3. I created other camera “Camera_Perspective2” that is more distant with some zoom in. Instead of Max Steps at 24, the difference in the problem is mainly visible at Max Steps 28. There is something with the distance. 4. I created an orthographic camera. The artifact problem between the cubes seems to be present at certain angles, but not the contrast in lower or higher density based on the camera distance from the volumes. **Tests:** There are 3 tests for this. - Set the viewport shading to “Rendered” to have the Cycles view. - Enable camera “Camera_Perspective”. Test 1. - Enable “Atmosphere_GN_Test_ProblemAt4MaxSteps” - Set Max Steps to 4. - Because the volume is not divided into many adjacent cubes, there are no artifacts visible between the cubes. However, at the bottom of the image (or closest part of the volume), we can see a higher density. It seems still related to the distance because moving the camera changes the transition area between the lower and higher density. Test 2. - Enable “Atmosphere_GN_Test_ProblemAt24MaxSteps” instead of “Atmosphere_GN_Test_ProblemAt4MaxSteps” - Set Max Steps to 24. - The same problem but at a higher Max Steps and with smaller adjacent volumes. Note: Changing the Max Steps to the default of 1024 and changing the Step Rate to the default value of 1.00 makes the artifacts visible again between the cubes. Changing the Step Rate to 0.1 makes the artifact disappear. Sadly, changing to 0.1 does not solve the problem for all scenarios. In my original project, I used this Step Rate and I had artifacts between the cubes for a cloud layer in high altitude. I also noticed that the problem was visible for density values like 0.02. To avoid the problem, I was limited to values a lot less dense like 0.0002. This might also explain why including VDB clouds in a volume with variable density also creates artifacts around the edges. Test 3. - Enable “Camera_Orthographic”. - Enable “Atmosphere_GN_Test_ProblemAt24MaxSteps”. - Rotate the camera using the parent object “Rotate” for the orthographic camera using the X axis. At 8 degrees, the artifacts start to become visible. Increasing the rotation brings the camera lower to the ground. - In the case of an orthographic camera, the angle is a factor to make the artifacts visible or not. **Conclusion** That seems to only happen for volumes of variable density. The initial project in this bug report highlighted the problem for overlapping volumes with variable density. This update highlights the aspect of the Max Steps with non-overlapping volumes. Furthermore, I reported this problem in another bug report under similar circumstances. Since then, it has been around a year and no follow-up has been done. Is this problem being investigated by the Blender team?

VolumeDensityBugWithMaxSteps.blend

1.6 MiB

Atmosphere_GN_Test_ProblemAt4MaxSteps.jpg

531 KiB

Atmosphere_GN_Test_ProblemAt24MaxSteps.jpg

557 KiB