Member-only story
Comparing and Explaining Diffusion Models in HuggingFace Diffusers
DDPM, Stable Diffusion, DALL·E-2, Imagen, Kandinsky 2, SDEdit, ControlNet, InstructPix2Pix, and more
Published in
33 min readAug 24, 2023
Table of Contents
- Introduction
- Prerequisites and Suggested Materials
- Diffusers Pipelines
- Pipeline: DDPM (Diffusion Models)
- Pipeline: Stable Diffusion Text-to-Image
- Pipeline: Stable Diffusion Image-to-Image (SDEdit)
- Pipeline: Stable Diffusion Image Variation
- Pipeline: Stable Diffusion Upscale
- Pipeline: Stable Diffusion Latent Upscale
- Pipeline: unCLIP (Karlo/DALL·E-2)
- Pipeline: DeepFloyd IF (Imagen)
- Pipeline: Kandinsky
- Pipeline: ControlNet
- Pipeline: Instruct Pix2Pix
- Appendix — CLIP
- Appendix — VQGAN
- Appendix — Prompt-to-Prompt
- Conclusions
- Acknowledgments
