songweige/rich-text-to-image — reverse-engineered prompt

Build me a local app that turns rich text prompts into images. I want a simple browser demo where I can type a prompt in a rich text editor and use formatting to control the result, like text color for object color, bigger text for stronger emphasis, different font styles for local art style, and footnotes or linked notes for extra details about a specific word or phrase. It should generate images from that formatted prompt and also let me paste or export the prompt as JSON so it can be reused from the command line.

Please make it work with the main Stable Diffusion models mentioned in the project, including the XL version and the anime XL option, with sensible defaults and a seed setting. Save outputs to folders, and keep the token or region visualizations available for debugging if possible. I also want the example prompts from the paper or demo to work so I can quickly test color control, local style control, and weighted emphasis. If anything is unclear, look up the current docs online and wire it up end to end.