Midjourney has launched Patchwork, its first storytelling tool that lets users create collaborative visual worlds.
During its weekly Office Hours meeting, Midjourney introduced Patchwork, a new tool that's immediately available to users. It's the first in a series of text and image generation tools designed to help users create visually consistent worlds, either individually or with others.
Patchwork is a collaborative, AI-supported infinite canvas for creating fictional worlds. It can help you go from vague story ideas to more complete ones; collaborate with other people to build a coherent setting together; and even assemble weird, collage-y visual stories from little scraps of images and text.
David Holz on Discord
Early access with some limitations
Max Kreminski, who leads Midjourney's Storytelling Lab, presented Patchwork in an X livestream. While it's the first tool ready for public use, Midjourney notes it's still in research preview and may have some bugs.
The tool features an infinite canvas similar to OpenAI's DALL-E 2 from 2022. Users can create "scraps" - small content pieces - using tools in a left sidebar. These tools generate elements across several categories: characters, events, locations, items, factions, and others. A language model assists with naming and writing descriptions.
Each canvas connects to others through "portals." When creating a new world, users provide a basic description and sample images to define the visual style. The system incorporates Midjourney's recent personalization features to maintain consistency.
The initial canvas displays the defining images and text description, which users can move around freely. These images serve as style references for all future content generation in that world.
While worlds are public by default, users can make them private or view-only. The system allows saving and loading different versions. Midjourney asks users to respect others' public creations and avoid damaging them.
Users interested in trying Patchwork can access it at patchwork.midjourney.com with a paid membership. The tool requires linking a Google account for login. Midjourney provides detailed setup instructions on their website.
The tool, though still in early development, could benefit various creative professionals, including tabletop gamers, authors, game designers, brand developers, and filmmakers.
Sharing and upcoming features
Several major updates are still pending from Midjourney's September announcements, including a 3D system, version 7 of their core technology, and the ability to generate eight images simultaneously. The company plans to release v7 after Christmas, possibly in January, which promises better prompt understanding. The video model comes later.