ComfyUI Node Runs on cloud
MOSS-Image Compile Prompt
Turn the caption or box layout a language model drafted, such as with the MOSS-Image prompt enhancer skill, into the prompt MOSS-Image reads. Layout boxes are placed on the image size being generated. The panel lists the caption and every box.
MOSS-Image Compile Prompt
- prompt
- caption
◄draft►
◄width1024►
◄height1024►
◄on_resizefit►
CategoryWAS Suite/Text
Inputs (4)
| Name | Type | Default | Description |
|---|---|---|---|
| draft | STRING | The model's answer holding `{"caption": "..."}` or a layout with `summary`, `items` and boxes in pixels. Words or code fences around the JSON are ignored; text with no JSON becomes the caption. | |
| width | INT | 102416–16384 | Width of the image being generated in pixels, such as `2560`. Layout boxes are placed on it. |
| height | INT | 102416–16384 | Height of the image being generated in pixels, such as `1440`. Layout boxes are placed on it. |
| on_resize | COMBO | fit | How a layout drafted at another size is placed: `fit` = scaled whole and centred; `fill` = scaled to cover; `stretch` = each side scaled; `keep size` = centred unscaled. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| prompt | STRING | The caption, or the layout as one line of JSON, which MOSS-Image reads as it is. Wire it straight into CLIP Text Encode with a MOSS-Image text encoder. |
| caption | STRING | The caption as plain text, also for a layout, for a preview or a node that reads plain text. |