camenduru/stable-diffusion-webui-colab

★ 15,942Jupyter NotebookAudience · generalComplexity · 1/5Setup · easy

Mindmap

mindmap
  root((sd-webui-colab))
    What it does
      Text to image
      DreamBooth training
      LoRA fine-tuning
      ControlNet guided gen
    Notebook tiers
      Lite stable
      Stable with ControlNet
      Nightly latest
    Requirements
      Google account
      Free Colab GPU
      No local install
    Use cases
      Quick image gen
      Model fine-tuning
      Art experimentation

mindmap root((sd-webui-colab)) What it does Text to image DreamBooth training LoRA fine-tuning ControlNet guided gen Notebook tiers Lite stable Stable with ControlNet Nightly latest Requirements Google account Free Colab GPU No local install Use cases Quick image gen Model fine-tuning Art experimentation

Click or tap to explore — scroll the page freely

Things people build with this

USE CASE 1

Generate AI images from text prompts using Stable Diffusion without owning a GPU or installing anything locally.

USE CASE 2

Fine-tune a Stable Diffusion model on your own photos using the DreamBooth or LoRA trainer notebook.

USE CASE 3

Use ControlNet to guide AI image generation with a reference sketch or pose for more precise results.

USE CASE 4

Save your Stable Diffusion setup to Google Drive so you can resume between sessions without re-downloading models.

Tech stack

Jupyter NotebookPythonStable Diffusion

Getting it running

Difficulty · easy Time to first run · 5min

This repository is no longer maintained, the author directs users to a newer replacement project.

In plain English

This repository provides a collection of ready-to-run Google Colab notebooks for Stable Diffusion, an AI image generation system that creates images from text descriptions. Google Colab (short for Colaboratory) is a free cloud service that lets you run code in your browser without installing anything on your own computer, using Google's servers. The notebooks here bundle together the Stable Diffusion web interface (a visual point-and-click UI for generating images) with various AI models, so anyone can start generating images by clicking an "Open in Colab" button and running the notebook. No local GPU or complex setup is required. The repository supports multiple modes and model variants. The readme mentions a DreamBooth and LoRA trainer branch, which lets you fine-tune an AI model on your own images, for example, teaching it to recognize your face or a specific art style. There is also a branch for saving the setup to Google Drive for persistence between sessions, and a ControlNet option, which is a tool for giving the AI more precise control over the structure and composition of generated images. The notebooks come in three stability tiers: "lite" (minimal and stable), "stable" (includes ControlNet), and "nightly" (latest features but potentially less reliable). The author notes in the readme that this repository is outdated and no longer receives updates, directing users to a newer project. Built as Jupyter Notebooks. The full README is longer than what was provided.

Copy-paste prompts

Prompt 1

I opened the stable-diffusion-webui-colab lite notebook and the web UI is running. Help me write an effective prompt and negative prompt to generate a photorealistic portrait.

Prompt 2

I want to use the DreamBooth notebook to train Stable Diffusion on 10 photos of my dog so it recognizes my dog as a subject. Walk me through the training steps and prompt format.

Prompt 3

I am using the ControlNet notebook. Show me how to upload a line drawing as a control image and generate a colored illustration that follows its structure.

Prompt 4

Help me save my model checkpoints and outputs to Google Drive from within this Colab notebook so they persist after the session ends.

Prompt 5

The nightly notebook is failing with a CUDA out-of-memory error. What settings in the web UI should I reduce to fit within the free Colab GPU memory limit?

Open on GitHub → Explain another repo

← camenduru on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.