ComfyUI Node
MOSS - Visual Regional Prompt
Experimental MOSS layout JSON using VRP bbox authoring and region chains. Connect optional CLIP to encode JSON positive and global negative conditioning, or use JSON/text outputs alone. Combined prompt joins scene and active item captions for a separate refiner encoder. Match generation dimensions to Canvas. Masks become enclosing boxes; no model patching or regional negative conditioning.
MOSS - Visual Regional Prompt
- canvas
- regions
- clip
- moss_prompt_json
- combined_prompt
- positive
- negative
◄vrp_state{"version":2,"canvas":{"width":1024,"height":1024},"regions":[]}►
◄mask_threshold0.50►
◄global_positive—►
◄negative_prompt—►
◄global_negative►
CategoryLuCiD VRP
Inputs (8)
| Name | Type | Default | Description |
|---|---|---|---|
| canvas | VRP_CANVAS | — | |
| vrp_state | STRING | {"version":2,"canvas":{"width":1024,"height":1024},"regions":[]} | — |
| mask_threshold | FLOAT | 0.500–0.99 | — |
| regionsopt | VRP_REGIONS | — | |
| global_positiveopt | STRING | — | |
| negative_promptopt | STRING | Optional negative text input; overrides the local field, including an empty string. | |
| clipopt | CLIP | — | |
| global_negativeopt | STRING | Local negative text. A connected negative_prompt input takes precedence. |
Outputs (4)
| Name | Type | Description |
|---|---|---|
| moss_prompt_json | STRING | — |
| combined_prompt | STRING | — |
| positive | CONDITIONING | — |
| negative | CONDITIONING | — |