🌁🪄🌃 - ControlNet with GPT-4

🌟 Born to Create: Controllable Text-to-Image Generation with GPT-4

Quick start: Colab Notebook ⚡

Hugging Face demo: ControlNet with GPT-4 🤗 (Need a GPU)

GPT-4 can generate code from a prompt, which can be rendered as an image, in a way that is true to the instructions to a higher degree of accuracy. This project is inspired by the paper - Sparks of Artificial General Intelligence: Early experiments with GPT-4.

P.S. You may need to have a GPT-4 API key since GPT-3.5 would not work properly. If you would like to run app.py locally, you need to specify torch==1.13.1 in requirements.txt.

Comparison of Stable Diffusion 2.1, DALL-E 3, and ControlNet with GPT-4

Prompt: a soccer ball to the right of a television and to the left of a cup, and they are all on a desk

Stable Diffusion 2.1:

DALL-E 3:

ControlNet with GPT-4:

Gradio Interface

Colab Interface:

Hugging Face Interface: