Video Depth AI is a Blender-native add-on from Blender Procedural that converts video footage into animated 3D scenes. AI models generate depth maps from your video, and those maps drive full 3D displacement to produce dynamic, animated geometry. It works with Blender 4.2 to 4.5 and supports both Cycles and Eevee render engines.
The workflow runs in a few steps: install and check dependencies in the addon panel, import footage using the "Import Video to Generate Depth" option, and let depth maps generate automatically. Clicking "Displace" then builds the animated 3D model. The AI can run locally or connect to external services depending on your settings, with local mode requiring an NVIDIA GPU. Note that the cloud option is not working properly at the moment.
A text-to-image generator is included as a bonus. Video Depth AI suits terrain and environment generation, depth-based art and experiments, and animation and VFX workflows. It's Blender Extension compatible and released under the GPL license.