I’m trying to set a more regular habit to learn and stay up to date with AI, in particular AI imaging. I’ve spent the past 9 months using Deforum to make a short film. And now want to dive into ControlNet, WarpFusion, and other tools. This isn’t meant for anyone else but purely for myself to collect thoughts and feel like I’m going in a direction.
Starting with WarpFusion… gosh dang I did not miss the early stages of figuring out how to run AI. I barely got it to pump out a video:
I also became incredibly inspired by this video. And now I want to center my learning on achieving this effect. Which will required full masking / flow understanding.
I learned:
One can run the google colab notebook locally! Which will be very helpful.
I will learn:
Conceptually why there is a GUI part of the notebook when the Discord says to not change it? How to use a model downloaded locally. Configure file management on my desktop so it doesn’t create a bunch of folders. Maybe use VSCode? In the further future I want to see how to generate my own mask videos.
Isaac Gazmararian 