Words belong in the picture.
Explore documented Chinese and English text rendering. Make language part of the composition, then proofread every output.
For the words that matter. And the worlds around them. Explore an open image model built with text and composition in mind.
Our creative ambition, not an accuracy guarantee. Hosted service in planning. Independent community site.
Posters, packaging, and the spaces in between. Original illustrative samples exploring what a well-placed word can do.
Illustrative samples / mockups — authored as SVG, not generated by MOSS-Image. Hover or focus a poster to read its concept prompt.
Explore documented Chinese and English text rendering. Make language part of the composition, then proofread every output.
Use experimental JSON layout prompts to describe positions for text and objects. More explicit input, more deliberate experiments.
Long prompts leave space for detailed art direction. Open Base and SFT weights give you two starting points to explore.
Enable JavaScript to explore the JSON layout presets.1024 × 1024 coordinate space · documented JSON fieldsModel details: official repository, model card, and user guide. The project’s reported capabilities have not been independently benchmarked here. See every source and what we could not verify.
From Fudan’s conversational MOSS to the OpenMOSS ecosystem and a new image model. A sourced history, with the important distinctions intact.
Read the storyA closer look at bilingual text, the real JSON layout format, long prompts, Base versus SFT, and the documented ComfyUI plugin.
Read the storyBuild a bilingual poster brief with clear hierarchy, exact copy, deliberate layout, and a repeatable review process.
Read the story