Report
freeTrialImage.bannerPity
freeTrialImage.upgradeUnlock
- ✓freeTrialImage.benefitHd
- ✓freeTrialImage.benefitWatermark
- ✓freeTrialImage.benefitUnlimited
Comfyui Moss Image
Comfyui Moss Image brings JSON layout prompts and bounding boxes into ComfyUI, with Base or SFT checkpoints and razor-sharp text rendering.
AI Tools
Explore AI tools
A Compact Layout Engine Built for ComfyUI Workflows
A 5B-parameter image family that reads structured JSON, keeps typography steady, and shifts easily between anime and photoreal looks.
- Bounding-Box Precision, NativelyEvery structured prompt entry carries a box, letting Comfyui Moss Image place objects, captions, and scene layers exactly where you define them.
- 4K-Token Prompts on a 5B CoreLong, detailed instructions up to 4K tokens run on a compact 5B backbone, with Base and SFT weights sharing one sampling interface.
- Readable Text in Any Art StyleTypography stays legible whether the output leans anime or photoreal, covering illustration, photography, and graphic design work.
Three Steps to a Finished Comfyui Moss Image Layout
From plugin install to a rendered layout in three quick steps.
Questions About Comfyui Moss Image
Answers to the things people ask most about running MOSS-Image layouts in ComfyUI.
Do I need the official ComfyUI repository?
Not at all — searching MOSS-Image in ComfyUI Manager installs the native plugin, while integration into the official ComfyUI repository is still in progress.
How does the SFT checkpoint differ from Base?
Both share one architecture and sampling interface; in the reference example, SFT keeps text grouping cleaner and the whiteboard composition tidier.
What role do bounding boxes play in a layout prompt?
Each box defines an element's position and size, and you can rewrite text, element descriptions, and boxes to rearrange both content and composition.
How long can prompts be, and which styles are covered?
Prompts run up to 4K tokens, with anime and photorealistic looks demonstrated across text, illustration, photography, and visual design.
Which memory modes suit low-VRAM machines?
Advanced settings expose GPU resident, Component, and Sequential modes; pick Sequential or shrink the image when memory is tight — Compile DiT is off by default and unavailable in Sequential.
Can the weights be fine-tuned under Apache-2.0?
Yes — the 5B backbone can be fine-tuned for custom tasks and domains, the weights ship under Apache-2.0, and pipeline code plus third-party components keep their own licenses.
Start Rendering Layouts with Comfyui Moss Image
Install the plugin from ComfyUI Manager, open the example workflow, and run a JSON layout prompt on Base or SFT weights.