sensity-ai/dot — reverse-engineered prompt

Build me a Python app called dot for authorized deepfake security demos. I want a simple desktop interface and a command line option where I can pick a source face image or a folder of images, choose my webcam as the target camera, select a built in face swap or image animation setup, and start a live preview that can be sent to a virtual camera.

It should work without training a new model each time, and it should let me switch source images while it is running. Include options for GPU acceleration when available, Apple Silicon support if possible, and a slower CPU mode for testing. Add clear warnings that this is only for research, demos, and approved penetration testing with consent.

Please include sample configs, a small demo data folder, setup instructions, model download instructions, and basic tests. Make it easy for someone to run from the GUI or with one command in the terminal.