A generative version of the Infrastructure CGI project. Comedy value only, not affiliated with InfraCGI
- A simple notebook runnable on Colab, showing use of Huggingface models to remove roads, vehicles and street furniture from images and replace it with imaginary infrastructure. Works on any web-based street view image, purely from text prompts with no visual editing or manual intervention.
I would love any contributions, whether interfaces to make it more accessible, improvements to prompts, or comedy value applications
Generative InfraCGI has its own Mastodon account and you are warmly invited to send pictures to it there and get back parallel responses to them.