https://github.com/outofai/stableenergy

A Simplified notebook for Smoothed Energy Guidance utilised for Stable Diffusion 2.1 base
https://github.com/outofai/stableenergy

attention-map classifier-free-guidance diffusion-model smoothed-energy-guidance stable-diffusion-2-1

Last synced: 4 months ago
JSON representation

A Simplified notebook for Smoothed Energy Guidance utilised for Stable Diffusion 2.1 base

• Host: GitHub
• URL: https://github.com/outofai/stableenergy
• Owner: OutofAi
• License: mit
• Created: 2024-08-20T20:13:51.000Z (10 months ago)
• Default Branch: main
• Last Pushed: 2025-01-10T00:05:34.000Z (5 months ago)
• Last Synced: 2025-01-10T01:19:05.344Z (5 months ago)
• Topics: attention-map, classifier-free-guidance, diffusion-model, smoothed-energy-guidance, stable-diffusion-2-1
• Language: Jupyter Notebook
• Homepage:
• Size: 17 MB
• Stars: 3
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# StableEnergy



[![Twitter](https://img.shields.io/twitter/url/https/twitter.com/cloudposse.svg?style=social&label=Out%20of%20AI)](https://twitter.com/OutofAi)

[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/OutofAi/StableEnergy/blob/main/SD_2_1_Smoothed_Energy_Guidance.ipynb)

A Simplified notebook for Smoothed Energy Guidance in Stable Diffusion 2.1 base

We put together a notebook implementation and explanation of Smoothed Energy Guidance based (SEG) of this paper https://arxiv.org/pdf/2408.00760

A query blurring method that yields significant improvements to the sampling process of SD 2.1 base and it's results in Vanilla and Classifier Free Guidance sampling.

| **Smoothed Energy Guidance for SD2.1 base** | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/OutofAi/StableEnergy/blob/main/SD_2_1_Smoothed_Energy_Guidance.ipynb) |

|----------------------|----------------------|

Some good results:

![image](https://github.com/user-attachments/assets/3e9e80a2-a5d3-4291-9319-ca71e6554757)

![image](https://github.com/user-attachments/assets/c5a9602f-b09d-4f36-94a4-3fd70462688a)