RedInk

Why we included this project

RedInk turns the usual Xiaohongshu publishing grind into one typed sentence: it drafts an outline, builds a cover, then renders each content page as a styled graphic. Because it pairs a text model with an image model, the copy and visuals stay coordinated and the text baked into the images renders legibly instead of coming out garbled. The flow is interactive, not a one-shot, so you can edit per-page descriptions, reorder cards, and regenerate any single page you dislike without redoing the batch. It runs as a self-hosted Flask backend with a Vue frontend and a ready-made Docker image, which means a small team can deploy it on its own hardware and plug in the API keys it already has. One caveat: it's released under a non-commercial CC BY-NC-SA license, so check the terms before building a paid product on it.