Skip to content
note · catap_art3d · 图像视频创作分享·· 29 days agoAI score55

Testing the MOSS-Image image generation workflow in ComfyUI

【最新画像生成AI】MOSS Imageを使ってみた話【ComfyUI】

AI briefing

The author tried the MOSS-Image text-to-image model open-sourced by Fudan's OpenMOSS team in the latest ComfyUI templates; the model comes in Base and SFT versions and supports three instruction forms: natural language, tag lists and JSON layouts. Generation takes about 30 seconds on an RTX3060 12GB, and the author also used an AI agent or Open WebUI with a local model to generate layout prompts, and said its Japanese text generation capability is close to Boogu-Image.

Open original note.com