Google Colab notebooks
Stable Diffusion with Diffusers
A powerful implementation of Stable Diffusion with a friendly web-based interface. It can be run locally on a computer with a decent graphics card or as a Google Colab notebook.
As well as text-to-image, image-to-image, inpainting, outpainting, upscaling and face restoration, it's capable of finetuning and blending custom models and much more.
Deforum has an emphasis on video generation. If you've seen videos or animations made with Stable Diffusion, this notebook was probably used.
Similar to Automatic1111, this has a friendly web-interface, text-to-image and image-to-image functionality and a lot more.
Possibly the simplest way of getting started with SD on Colab, but also the most basic. Great for trying it out, but use something more fully-featured if possible.
It still blows my mind that it's possible to run Stable Diffusion locally on my Macbook Air, taking ~15 seconds for a 20 step generation. Models need to be converted to Core ML before use, although a few ready-made ones can be downloaded. While it's fast, it doesn't currently have image to imgage, inpainting or any other advanced features.
This app also uses Apple's Neural Engine for fast generation, though it's still in beta and can only be installed via Testpilot.
DiffusionBee only uses the GPU so generation takes between 1-3 mins for 25 steps on an Air. However, it's pretty fully featured with image to image, inpainting and outpainting.
Even more astonishing than running SD on a Macbook, this is a fully-featured app for running it on an iPhone or iPad. It's pretty slow on my 14 Plus, but I'd imagine it would work well on an iPad Pro. But it's impressive it's possible at all.
All-in-one app for running SD locally. Similar to DiffusionBee.
Command-line interface that includes prompt-based image editing and masking.
A web app and API from Stability AI, the company behind Stable Diffusion, this is a great way to demonstrate what SD can do. It's extremely fast – images are generated in seconds – but the interface can sometimes be confusing and it doesn't have all the features of something like Automatic1111.
A combined text and image AI-powered storyteller, NovelAI can write stories, generate interactive fiction and much more.
NightCafe has been around for a long time in the AI art space, starting as a web-based VQGAN+CLIP generator and rapidly adding SD when it was released.
Thanks for reading! If you know of any other apps, notebooks or services I've missed, please do let me know.