GPT-3 Guided Art

I’ve been experimenting with GPT-3 as providing captions that are then fed into a CLIP-based neural nets. Here is an example from GPT-3 —> VQGAN+CLIP. The phrase in quotes was the prompt: “Photorealistic image of” a man with a beard and a hat.
This is a great photo of a man with a beard and a hat. The man is looking straight at the camera and smiling. The photo is in black and white and the background is a dark blue.


Here is a movie of the iterative process: GPT3ART: Man with Beard & Hat - YouTube