Dreambooth - Diffusion Stash

Dreambooth

One of the most promising techniques in the Stable Diffusion world is known as «Dreambooth». It allows you to teach Stable Diffusion about very specific concepts with just 5-20 pictures of that thing. Here's how it works: you feed it a small number of images about one single thing (an object, a person... anything identifiable and unique), and it will make Stable Diffusion an expert in that particular thing. Then, you can use the model to generate very high quality thing-specific images, like custom avatars if you've trained it to learn about your face – or pictures of yourself in Paris even if you've never been there.

Dreambooth Stable Diffusion

Implementation of Dreambooth by way of Textual Inversion for Stable Diffusion. Tweaks focused on training faces, objects, and styles.

Liner

Free tool that lets you train your ML models easily. It takes your training data and gives you an easy-to-integrate ML model. No coding or expertise in machine learning required.

Stable Diffusion Dreambooth Models

Browser for the HuggingFace DreamBooth library.

SD Textual Inversion Embeddings

Dreambooth Gui by smy20011

(10GB VRAM) Easiest to use with a nice Web UI.

dreambooth