MOSSImageLayoutPrompt - ComfyUI Built-in Node Documentation

Builds the layout prompt that MOSS-Image reads. It takes the elements drawn in the Create Bounding Boxes node and emits a JSON layout with one entry per box: its label, its caption, and its rectangle in the pixel space of the target canvas. Set width and height to the size of the image you plan to generate, then connect the output to the prompt input of the text encoder. With no elements connected the prompt text is returned unchanged.

Inputs

Parameter	Description	Data Type	Required	Range
`prompt`	Description of the whole image. It becomes the caption of the layout.	STRING	Yes	-
`element`	The elements to place, taken from the Create Bounding Boxes node. Each box contributes its caption; text boxes also contribute their text, and boxes that carry a color palette keep it.	ARRAY	Yes	-
`width`	Width of the generated image in pixels. The boxes are converted into this pixel grid, so use the size of the image you plan to generate (default: 1024).	INT	Yes	16 to 16384 (step 16)
`height`	Height of the generated image in pixels. The boxes are converted into this pixel grid, so use the size of the image you plan to generate (default: 1024).	INT	Yes	16 to 16384 (step 16)
`character_boxes`	Numbers of the boxes to label as characters instead of objects, as tagged in the Create Bounding Boxes preview, for example “1, 3”. Every number in the string is read, so any separator works (default: empty).	STRING	Yes	-

Outputs

Output Name	Description	Data Type
`prompt`	The layout prompt: a JSON object holding the caption, one entry per box, and the box coordinate format. The input text is passed through when no elements are connected.	STRING

Notes

Box rectangles arrive from Create Bounding Boxes as normalized coordinates between 0 and 1000 and are written to the layout in pixels relative to the center of the canvas, in yxyx order. The width and height values therefore have to match the image you generate, otherwise the boxes land in the wrong place. A box that holds text is always labeled as text, a box whose number is listed in character_boxes is labeled as a character, and every other box is labeled as an object.

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): da012a8d0dceaf6391110c90bd677d7929920d60fe961c00e5011b037206aa9a

​Inputs

​Outputs

​Notes

Inputs

Outputs

Notes