AliaksandrSiarohin/first-order-model ? reverse-engineered prompt

Build me a working version of this project for animating a still image with motion from a driving video, like in the README. I want to be able to install it, run the demo with a source image, a driving video, and a checkpoint, and get a result video out without having to figure out the setup myself.

Please make the repo easy to use on a fresh machine, including clear setup steps, sensible defaults, and any fixes needed so the demo and training scripts run smoothly. If anything depends on extra data, checkpoints, or other tools like ffmpeg or Docker, make that obvious and wired up in a simple way. It would also be great if the notebook demo and the command line demo both work, and if there’s a clean way to run training or reconstruction from the existing config files.

Look up current docs online if you need to, and make the smallest changes needed to get the main workflow working end to end.