Report
freeTrialImage.bannerPity
freeTrialImage.upgradeUnlock
- ✓freeTrialImage.benefitHd
- ✓freeTrialImage.benefitWatermark
- ✓freeTrialImage.benefitUnlimited
Comfyui Moss Image
Comfyui Moss Image gives ComfyUI creators JSON bounding-box control, Base and SFT checkpoints, 4K-token prompts, and sharp on-image text in one plugin.
AI Tools
Explore AI tools
How Comfyui Moss Image Powers Layout-Driven ComfyUI Renders
This 5B text-to-image model combines JSON-driven composition with reliable text rendering, covering anime illustration and photorealistic scenes.
- JSON Prompts That Place Every ElementStructured JSON prompts use bounding boxes to define where objects, captions, and scene details appear and how large they should be.
- Long Prompts on a Lean 5B ModelFeed prompts up to 4K tokens into a compact 5B backbone; Base and SFT weights share the same sampling workflow.
- Readable Text Across Art StylesGet clear on-image typography alongside flexible anime and photoreal aesthetics for illustration, photo, and design work.
Run Comfyui Moss Image in Three Steps
Go from plugin setup to a finished JSON layout render without leaving ComfyUI.
Comfyui Moss Image Questions, Answered
Quick answers for creators setting up MOSS-Image layout workflows in ComfyUI.
Do I need the official ComfyUI repository?
Not yet. You can install the native plugin by searching MOSS-Image in ComfyUI Manager; official repository integration is still underway.
How is MOSS-Image-Base different from SFT?
They use the same architecture and sampling controls, but the sample shows SFT producing tighter text grouping and a cleaner whiteboard arrangement.
What role do bounding boxes play in a layout prompt?
Every box defines an item's location and dimensions; edit the text, element notes, or box coordinates to reshape both content and composition.
What prompt length and styles can I use?
You can work with prompts up to 4K tokens, and the model covers anime and photoreal looks for typography, illustration, photos, and design.
Which memory modes suit low-VRAM GPUs?
Advanced options include GPU resident, Component, and Sequential modes. Pick Sequential or lower the image size when memory is limited; Compile DiT stays off by default and is not available in Sequential.
Can I fine-tune the weights under Apache-2.0?
Yes. The 5B backbone can be adapted to new tasks and domains, and the weights are Apache-2.0 licensed; pipeline code and third-party parts retain their own terms.
Start Building Layouts with Comfyui Moss Image
Add the plugin through ComfyUI Manager, load the sample workflow, and test a JSON layout prompt using either Base or SFT weights.