AI GitHub
ComfyUI for node-based image generation
Build reusable generation graphs around models, samplers and conditioning inputs.
Key Takeaways
- ComfyUI nodes pass data such as latents, masks and conditioning through a graph.
- Start with a graph using known nodes and compatible model artifacts.
- Open a small documented workflow, pin its seed and replace one conditioning input.
- A shared graph can depend on custom nodes and local models.
How it works
ComfyUI nodes pass data such as latents, masks and conditioning through a graph. Saving that graph makes the generation steps visible and reusable.
What you need
Start with a graph using known nodes and compatible model artifacts. Store the original graph before adding a custom node so a failure can be isolated.
What to try first
Open a small documented workflow, pin its seed and replace one conditioning input. Save the workflow and the checkpoint identifiers with the output.
Use a graph input that changes only one region. Compare the workflow JSON and generated result so you can reproduce the edit without relying on the final image alone.
Where to look for failures
A shared graph can depend on custom nodes and local models. Review code before installing nodes and record their versions before sharing a recipe.
Related Franklin pages
Start with a small workflow that can be rerun. Keep the seed, mask, model revision and graph together so the visual output can be compared after an edit.
- Generate and revise images through a canvas and reusable workflows. InvokeAI for image generation and canvas editing
- Generate animations by combining compatible image-model components with motion modules. AnimateDiff for adding motion to image diffusion workflows
Franklin’s image tests give you examples of outputs to compare across visual models. Franklin’s image tests
Project documentation
Try it for your workflow
Explore the project’s documentation for setup and usage, or save this page to a collection for later.
Read the project documentation ↗