PROJECT / ABOUT
About Image2
Image2 Studio is a local image-task workspace for OpenAI-compatible gateways. It turns a Base URL, API key, and gpt-image-2 into a desktop workflow that can plan, queue, edit, and trace every image version.
Why Image2 exists
OpenAI-compatible gateways make model access flexible, but general chat clients rarely manage image-reference roles, regional edits, paid-call ordering, and parent-child versions together. Image2 gives those image-specific structures one quiet desktop surface.
From gateway to image task
You configure an HTTPS Base URL, API key, agent protocol, and model names. The agent interprets a goal and splits it into independent image tasks; the desktop app executes them in a strict serial queue so stop, resume, retry, and cost-control states stay visible. References can represent identity, product, pose, composition, material, or style, while region annotations become structured edit requests.
An explicit data boundary
The API key is stored in macOS Keychain or Windows Credential Manager. Conversations, tasks, images, and annotations stay on the device. Generation and edit requests are sent to the gateway you configured and are subject to that gateway's and its upstream providers' terms. The marketing site is separate from the desktop workspace: it does not accept keys or read local project content.
Developed in public
Image2 source, releases, and development history are public on GitHub. You can inspect the implementation, download releases, report reproducible issues, or contribute improvements through pull requests. Include the environment and reproduction steps needed to investigate a problem, but never publish API keys, access tokens, or sensitive logs.
Contact the project
Feature proposals, bug reports, and documentation issues are tracked through GitHub Issues so discussion, status, and fixes remain public and traceable.