hero8152/Infinite-Canvas — reverse-engineered prompt

Build me an Infinite Canvas style desktop web app for AI creation workflows. I want a big canvas where I can organize prompts, images, videos, and workflow steps, then run them through different AI services from one place.

It should let me connect to OpenAI compatible APIs, Gemini style APIs, ModelScope models, RunningHub workflows, Volcengine, Jimeng CLI, and a local ComfyUI server on my network. I want simple settings screens for keys, base URLs, model choices, and service login where needed.

The app should support text to image, image to image, text to video, image to video, expanding images, previewing 360 panorama screenshots, extracting video frames, and reusable loop style workflow nodes. Make it friendly for beginners, with clear buttons, status messages, saved outputs, and useful error messages.

Please include easy start scripts for Windows and Mac, and look up current docs online if you need to.