Report
freeTrialImage.bannerPity
freeTrialImage.upgradeUnlock
- ✓freeTrialImage.benefitHd
- ✓freeTrialImage.benefitWatermark
- ✓freeTrialImage.benefitUnlimited
Comfyui Moss Image
Generate precise ComfyUI layouts with Comfyui Moss Image — JSON bounding-box prompts, Base or SFT checkpoints, and razor-sharp on-image text.
AI Tools
Explore AI tools

A Closer Look at the Comfyui Moss Image Layout Engine
Structured JSON prompts meet a compact 5B model, so elements land exactly where you place them and rendered text stays clean.
- Bounding Boxes That Actually StickGive every object a position and a size in structured JSON, and the renderer respects those coordinates for shapes, captions, and scene elements.
- Long Prompts, Lean HardwarePrompts can stretch to 4K tokens while a 5B backbone keeps resource use modest; Base and SFT checkpoints plug into the same sampling interface.
- Legible Text Across Art StylesTypography stays readable whether you render anime illustrations, photographic scenes, or design mockups, with both looks sharing one workflow.
Get Comfyui Moss Image Working in Three Steps
From adding the node to exporting a finished layout, here is the short path.
MOSS-Image in ComfyUI: Common Questions
Quick answers on installation, checkpoints, bounding boxes, memory modes, and licensing.
Do I need the official ComfyUI repository to use it?
Not at all. You install the native plugin by searching MOSS-Image inside ComfyUI Manager, while work on merging it into the official repository is still underway.
How is the SFT checkpoint different from Base?
They share one architecture and sampling interface. In the reference example, SFT keeps text groups cleaner and produces a neater whiteboard composition.
What role do bounding boxes play in a layout prompt?
Each box pins an element's position and size. You can rewrite captions, adjust element descriptions, and move boxes to reshape both content and composition.
How long can prompts be, and which styles work best?
Up to 4K tokens per prompt, with anime and photorealistic looks demonstrated across text blocks, illustrations, photography, and visual design.
My GPU has little VRAM — which mode should I pick?
Advanced settings expose GPU resident, Component, and Sequential modes. Choose Sequential or shrink the image when memory is tight; Compile DiT stays off by default and cannot run in Sequential.
Am I allowed to fine-tune the weights?
Yes. The 5B backbone ships under Apache-2.0, so you can adapt it to your own tasks and domains; the pipeline code and third-party components carry their own licenses.
Put Comfyui Moss Image to Work Today
Grab the plugin from ComfyUI Manager, load the sample workflow, and push your first bounding-box JSON prompt through Base or SFT.


