We use Diffusers to re-train a better depth-conditioned ControlNet based on our Depth Anything.
Please download our config file and pre-trained weights, then follow the instructions in Diffusers for inference.
Please refer to our project page. We use MagicEdit to show demos of video editing based on depth information.