I have four poodles: Bella, Estella, Brave and Blessy. I wanted to turn them into animated characters and tell little stories about their everyday lives. They have similar coats, and I know their faces so well that I thought a few good photos would be enough. Then I looked at one of the images I’d made of Bella and realized she had Blessy’s muzzle.
I take new photos of them every so often. Sometimes I put the project aside for a while. When I come back, I remember finally getting a dog to look right, but the AI sometimes picks up from a different picture. I end up scrolling through our old chats looking for the one where I said, “Yes, that’s him.”
I tried making reference sheets for each dog. When I finally got Brave looking like himself, I asked for the same image in a different layout. He came back shorter. I already had the version I wanted. I just needed a way to say, “Use this one. I like his face and his height. I’m only changing the layout.” My folder was still small enough to go back and find that picture. I kept thinking about how much harder that would get after a few months of making stories.
That’s why I wrote Hush Production Brief. Before I start a new session, I want to ask for Brave and get a short note pointing me back to the picture I chose, what I liked about it, and anything I still haven’t decided. I wrote those steps as a Skill for Uliya and checked them locally with made-up examples. I still need to try it on a real NAS. What I want is to come back to a story and pick up where I left off with the same four dogs.
