TencentARC/GFPGAN ? reverse-engineered prompt

Build me a Python app that can take a blurry or damaged face photo and restore it into a cleaner, more natural looking face.

I want to be able to point it at one image or a whole folder, then save the restored results into an output folder. It should have a simple command line entry point for inference, and support choosing between a few model versions, plus an option to upscale the final image. If possible, let it also enhance the background so the whole picture looks better, not just the face.

Please make it easy to install and run, with clear setup instructions and a small demo input folder so I can test it quickly. If you need to check current docs for model or dependency details, go ahead and look them up online. Include a clean version that works without special custom build steps if you can.