https://github.com/mpaepper/stablediffusion_magicmix
Implementation of the ByteDance MagicMix paper
https://github.com/mpaepper/stablediffusion_magicmix
deep-learning generative-ai stable-diffusion
Last synced: about 1 year ago
JSON representation
Implementation of the ByteDance MagicMix paper
- Host: GitHub
- URL: https://github.com/mpaepper/stablediffusion_magicmix
- Owner: mpaepper
- Created: 2022-11-04T01:06:36.000Z (over 3 years ago)
- Default Branch: main
- Last Pushed: 2022-11-04T01:56:42.000Z (over 3 years ago)
- Last Synced: 2025-03-29T12:02:42.735Z (over 1 year ago)
- Topics: deep-learning, generative-ai, stable-diffusion
- Language: Jupyter Notebook
- Homepage: https://www.paepper.com/blog/posts/everything-you-need-to-know-about-stable-diffusion/
- Size: 3.93 MB
- Stars: 20
- Watchers: 2
- Forks: 1
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# StableDiffusion MagicMix
Since the ByteDance paper titled "MagicMix: Semantic Mixing with Diffusion Models" (https://arxiv.org/abs/2210.16056) didn't publish their code, I've implemented a Jupyter notebook here, so you can try it out.
The notebook implements a function called `magic_mix` which takes the path to an image and the prompt towards which it should adapt the image.
Additional optional parameters:
nu: controls how much the prompt should overwrite the original image in the initial layout phase. If your result is too close to the original image, try increasing this parameter.
total_steps: number of inference steps for stable diffusion
guidance_scale: this is the classifier free guidance. The higher this is set, the more it will drive your result towards your prompt.
Examples:
