Feedback
freeTrialImage.bannerPity
freeTrialImage.upgradeUnlock
- ✓freeTrialImage.benefitHd
- ✓freeTrialImage.benefitWatermark
- ✓freeTrialImage.benefitUnlimited
MOSS-Image
MOSS-Image turns detailed briefs and structured JSON into controlled renders, so you can position every object, caption and bounding box with precision.
AI Tools
Explore AI tools
Free AI Image
Truly Free AI Image Generator
Image Upscaler
Enhance image resolution effortlessly
AI Color Changer
Transform photo colors in seconds
AI Room Designer
Design your dream room with AI-powered 3D rendering
Skin Enhancer
Enhance skin texture naturally with AI
Face Editor
Enhance face texture naturally with AI
AI Clothes Changer
The Advantages of MOSS-Image
A five-billion-parameter image model that treats layout instructions as first-class input.
- Direct Layout GuidanceAssign a dedicated region to every object and text label, so the composition follows your JSON rather than model guesswork.
- Five-Billion-Parameter Core, Long ContextExtended prompts remain readable, while the same weights can be fine-tuned for specialized tasks and domains.
- Two Checkpoints, One ArchitectureBoth releases use Apache-2.0 licensing and the same structure; the SFT version produces cleaner whiteboard text.
Launch MOSS-Image in Three Steps
Go from Python setup to a precisely positioned image in three moves.
Questions About MOSS-Image
Answers on prompts, model checkpoints and installation for MOSS-Image.
What can I define with layout control?
Your JSON describes the overall scene, individual objects and visible text, and assigns each element a position-and-size box.
What separates MOSS-Image-Base from MOSS-Image-SFT?
Both use the same architecture and sampling API. The SFT release is tuned on curated data and, in the project demo, delivers more coherent text grouping and a neater whiteboard arrangement.
What is the maximum prompt length for MOSS-Image?
It accepts as many as 4K tokens, which is ample for a multi-element layout JSON.
Which hardware and software are required?
You need Python 3.10 or newer plus an NVIDIA GPU and a compatible PyTorch build. When VRAM is constrained, use Sequential mode or lower the image size; Compile DiT stays off by default and cannot be used in that mode.
Is ComfyUI supported for MOSS-Image?
Yes. Find MOSS-Image in ComfyUI Manager, install the native plugin and open the sample workflow; official ComfyUI repository integration is still underway.
Which license covers the weights and code?
The model weights use Apache-2.0. Pipeline code and third-party components retain their respective licenses; cite the project repository for research work.
Put MOSS-Image to Work on Your Next Layout
Set up the Diffusers pipeline or ComfyUI plugin, test a structured layout prompt, or refine your concept with the MOSS-Image Prompt Enhancer.