Using Depth Maps to Create Semi-Volumetric Video in Blender
In this video, I go through my full workflow for turning regular video into semi-volumetric 3D inside Blender using depth maps.
I start with two ways of generating depth: DaVinci Resolve’s AI Depth Map and Video Depth Anything, an open-source video depth-estimation model designed to produce temporally consistent depth across video. I also show how I run Video Depth Anything locally through ComfyUI.
From there, I bring the original footage and depth map into Blender, where the depth information is used to displace geometry with Geometry Nodes and create the semi-volumetric effect.
The depth map itself is only data - once it's inside Blender, it can be used to control position, scale, particles, instances, color, density, and much more.
Resources
Video Depth Anything — Official GitHub
CVPR 2025 Highlight: Video Depth Anything: Consistent Depth Estimation for Super-Long Videos
DaVinci Resolve
AI Depth Map / Depth Map v2 is available in DaVinci Resolve Studio.
Music
Lovers Entwined, Hainbach
Youtube Audio Library
Videos downloaded from Pexels:
Mohammad Hedayet Sarker
Jeeva K Vinoy
I'm continuing to experiment with different ways of working with spatial and volumetric video - including depth maps, RGBD capture, point clouds, Gaussian splatting, and other techniques for bringing real-world footage into 3D.
Follow along on Instagram for shorter experiments and new techniques:
@p_h_r_o_