In this article, we’ll explore how to build cinematic cloudscapes using our Cloud Formation Pack 2 assets. Using 3ds Max and V-Ray as our examples, we’ll cover practical techniques for placing, layering, shading, and rendering volumetric clouds.
The same principles can be applied to your preferred DCC, so you can adapt the workflow to fit your own setup.
To get started, we’ve prepared a camera, an airplane, and a Dome Light with an HDRI. Our goal is to build a cloudscape around the aircraft. For the lighting, we’ve rotated the HDRI to create a backlit setup, giving the clouds more depth and definition. Backlighting makes clouds really beautiful, so it’s a quick way to increase the realism and quality of your render.
Depending on your DCC, the workflow may vary. Here, we’ll use 3ds Max and V-Ray as an example. To load the cloud formation asset, we use V-RayVolumeGrid. In the Input rollout, select `cloud_201_002.vdb`. V-Ray automatically recognizes the sequence and displays it as `cloud_201_####.vdb`. Set Mode to Cache Index and set Index to 2 to load the specific asset you want.
One of the advantages of using VDBs as volumetric assets is their flexibility. You can freely move, rotate, and scale each cloud in the scene. Each Volume Grid can also use a different cloud asset. In this example, I’m using cloud_formation_002, cloud_formation_002, and cloud_formation_004 from right to left.
Since Cloud Formation Pack 2 contains multiple cloud formations in each asset, you only need a few VRayVolumeGrids to fill the camera’s view. As the viewing area becomes wider with distance, the cloud volume needs to cover a larger area. For this reason, we scale the Volume Grids progressively, with the farthest one being the largest. This creates a continuous cloudscape while keeping the number of assets to a minimum.
Depending on the shot, you can mix and match different cloud formations to create a wide range of cinematic looks and compositions. For even more variety and detail, layer in other assets from our cloud collection, such as cumulus and anvil clouds, to build richer and more dynamic skies.
Clouds are made up of water droplets and ice particles that scatter light in complex ways. In CG, we can approximate this behavior by adjusting the Phase Function in V-Ray, which controls the direction in which light scatters inside the volume. In other rendering packages, this parameter may be called Anisotropy or have a different name.
From left to right: Phase Function set to 0.7, 0.0, and -0.7.
A positive Phase Function value produces stronger forward scattering, directing more light toward the camera and creating brighter, more defined clouds. Here, we compare -0.7, 0.0 (default), and 0.7 to clearly demonstrate the difference.
The positive value produces brighter highlights and greater contrast, while the default and negative values result in darker, flatter-looking clouds. In this particular setup, 0.7 produces the most realistic-looking result.
From left to right: Phase Function set to 0.7, 0.0, and -0.7.
Let’s rotate the HDRI by 180° to create a front-lit setup and run the same comparison. Again, 0.7 produces the most convincing result in this setup, while the other values make the clouds appear either flatter or overly dense.
Now that we’ve covered shading, let’s move on to rendering. Volumetric assets contain a large amount of data, which can lead to long render times. Rendering optimization is especially important for animation, where even small savings can add up across hundreds of frames.
In V-Ray, the Probabilistic Volumetrics Direct Samples setting can help optimize volumetric rendering. The default value is 20, but you can adjust it to reduce render times while maintaining acceptable image quality.
Note that other renderers may not have a direct equivalent to this setting. In Redshift, however, settings like Volume Samples, Unified Sampling, Volume Trace Depth, and Extinction Falloff give different ways to balance volumetric rendering quality and performance. These settings can be adjusted to find a nice balance between render time and image quality.
From left to right: Direct Samples set to 20 (default), 10, 5, and 3, with render times of 12m 28s, 11m 02s, 10m 54s, and 11m 08s, respectively.
We tested four values for Probabilistic Volumetrics Direct Samples. There is no visible difference in image quality between the settings, while a value of 5 delivers the fastest render time in this test. A value of 5 reduces render time by approximately 12.6% compared with the default value of 20.
That’s it for the rendering setup. Next, let’s look at a few more examples using the Cloud Formation Asset Pack 2.
In this shot, three F-35s fly in formation. We use just two cloud assets: cloud_201_002.vdb for the foreground and cloud_212_004.vdb for the background. The scene is lit with V-Ray Sun and Sky.
Top view of the F-35 shot, showing the two VRay Volume Grids positioned and scaled to cover the camera frame.
Top view of the Pan Am airliner shot. The three V-Ray Volume Grids have been rotated and scaled to fill the camera view.
Here’s the final example: another airliner close-up using cloud_201_004.vdb. As you can see, composition and scale can dramatically change the final look, even when using the same cloud asset.
Once again, we duplicated the asset three times and scaled each instance to fit the camera view.
One final tip: I added a large white plane to bounce light onto the Boeing 767, helping it stand out in the final render. This is especially useful for close-up shots.
I hope these tips help you create better above-the-cloud shots with our assets. See you next time!
Also, if you need any help with custom simulations, please e-mail us. We have a great team who eat, sleep, and breathe this stuff. We would love to work with you!
Want to learn more about VDBs? Check out this tutorial: Your Clouds Suck. How to Light and Shade VDB Clouds for Realistic Renders!