How to Use Aesthetic Gradients - Stable Diffusion Tutorial

Learn how to use Aesthetic Gradients in Stable Diffusion with this comprehensive 40-minute tutorial. Explore the recent paper on personalizing text-to-image generation through aesthetic embeddings. Follow step-by-step instructions for installing the necessary tools, using pretrained embeddings, and training new ones. Compare different embeddings and analyze experiment outcomes. Gain insights into enhancing image generation by specifying preferences more clearly with existing Stable Diffusion models. Suitable for users with Python 3.10, a local Nvidia GPU, and CUDA 11.3+.

- Summary
- Paper Explanation
- Webui Installation
- Aesthetic Gradients Installation
- Using Pretrained Embeddings
- Training New Embedding
- Comparing Embeddings
- Experiment Outcomes
ffmpeg resize and crop video: ffmpeg -i ghib.mp4 -c:a copy -filter:v "scale=92,crop=iw-448:ih-0" smol-ghib.mp4