paper ControlNet

[junxnone]

ControlNet

• Control diffusion models
• 添加输入到 Stable Diffusion 中 控制生成内容
  • Canny Edge
  • M-LSD Lines
  • HED Boundary
  • User Scribbles
  • Fake Scribbles
  • Human Pose
  • Semantic Segmentation
  • Normal Map
  • Anime Line Drawing

---

Input Demo

Input | Demo -- | -- Canny Edge |  M-LSD Lines |  HED Boundary |  User Scribbles |  Fake Scribbles |  Human Pose |  Semantic Segmentation |  Depth |  Normal Map |  Anime Line Drawing | 

Arch

Reference

• paper 2023 ControlNet Adding Conditional Control to Text-to-Image Diffusion Models [Code]