I implemented the Stable Diffusion engine logic as a script (rather than the usual notebooks) so that you can call the script with various command-line parameters. Then for good measure, added a callback that can be passed to the engine which would (at least currently) save image frames at given intervals.
I’m not quite happy with the callback implementation since it’s not generic enough and simply saves frames at the moment, but the plan was to expand upon it and maybe even add an option to save videos using generated frames from within the script.
But then I went off on a tangent since I wanted to see if I could implement the same logic using Keras/Tensorflow since I work/develop on a Mac and Keras/Tensorflow has been generally more stable and more responsive.
So I added another Stable Diffusion engine (this time using Keras/Tensorflow) and added the ability to switch between engines using a command-line switch 
That’s where it is at the moment though I did want to add some more features (negative prompts, img2img etc.) to both engines, but not sure if I’ll manage that this week or not …
Here’s the Git repo with the code: