Because one company alone cannot keep up with the pace of AI development, open-source developers are expected to help.
AI startup Stability AI has released an open-source version of its text-to-image platform DreamStudio under the name StableStudio. The goal is to create a “world class user interface for generative AI which users fully control”. From the beginning, DreamStudio has been the primary interface for Stable AI’s latest models and features. Recently, SDXL, Stable Diffusion’s latest and greatest model, was added.
With StableStudio, the company is now bringing the interface from the cloud to edge with a few tweaks.
StableStudio currently only runs with Stability API
To support third-party developers as much as possible, Stability AI has been experimenting with a new plugin system. StableStudio still uses the Stability API for inference, but future plans include local inference via WebGPU and Automatic1111, as well as support for features like ControlNet. Bounties will be announced for certain features to provide financial incentives for open-source developers.
DreamStudio was originally developed for animation using Disco Diffusion, but has evolved into image generation with the release of Stable Diffusion in the summer of 2022. However, the goal of creating an “a great multi-modal experience for generative AI” has not been abandoned, according to the announcement.
Chat integration planned with StableVicuna
The company will continue to operate DreamStudio as a commercial platform while developing StableStudio. In addition to further image improvements, the company also plans to release a chat interface based on the StableVicuna language model for StableStudio.
Stability AI’s StableStudio is available on GitHub. Users can also leave feedback and requests there.