Our team works on short films and sequence shots, and the biggest headache with ComfyUI has never been image quality—it’s iteration management. I can generate a single shot fifteen times, and one of those versions will be perfect, but two days later I have zero clue which seed or which image it came from. Re-generating overwrites stuff, the history tags are a total mess, and stitching shots into a sequence turns into a graveyard of files. So I built myself a tool, and it kinda snowballed from there.
The core idea is to lock down three things every time you generate: the input, the corresponding workflow, and the output. It uses a moodboard-like interface—start from an input frame, hook it up to a workflow, and you can open it anytime in your local ComfyUI with params synced automatically. Collaboration gets smoother too: you share the whole canvas, and your teammates get a pipeline they can run directly, no more asking who’s connected to what. Planning to open-source it, it’s super rough, so bring on the criticism.