ComfyUI Node
Krea 2 / MOSS Image Prompt Builder
A ComfyUI node in text/prompt with 14 inputs and 2 outputs.
Krea 2 / MOSS Image Prompt Builder
- clip
- image
- prompt
- negative_prompt
◄target_modelKrea 2►
◄prompt►
◄style_override►
◄detailStandard►
◄use_imagefalse►
◄use_wildcardfalse►
◄wildcard_items►
◄wildcard_pick0►
◄seed0►
◄temperature0.70►
◄max_tokens768►
◄thinkingfalse►
Categorytext/prompt
Inputs (14)
| Name | Type | Default | Description |
|---|---|---|---|
| clip | CLIP | An LLM-type text encoder that can generate text (Qwen3.5, Qwen3-VL, Gemma...). Needs vision for image mode. | |
| target_model | COMBO | Krea 2 | Which model's prompt format to write. |
| prompt | STRING | Your basic idea. In image mode it becomes optional extra direction; in wildcard mode it is a core idea to keep. | |
| style_override | STRING | Optional. If filled, the prompt is forced into this style in every mode. In image mode the image's own style is ignored. e.g. 'watercolor illustration'. | |
| detail | COMBO | Standard | 3 options: Concise, Standard, Detailed |
| use_image | BOOLEAN | false | Describe the connected image as a prompt in the target format. |
| use_wildcard | BOOLEAN | false | Invent a scene from the ingredient list below. |
| wildcard_items | STRING | Ingredients in any order, one per line (or comma separated). Leave empty for a fully random scene. | |
| wildcard_pick | INT | 00–64 | 0 = use every ingredient. N = randomly pick N of them each run. |
| seed | INT | 00–18446744073709550000 | — |
| temperature | FLOAT | 0.700.05–1.5 | — |
| max_tokens | INT | 76864–4096 | Raise this if you turn thinking on. |
| thinking | BOOLEAN | false | Let the LLM reason first (if the model supports it). Slower. |
| imageopt | IMAGE | — |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| prompt | STRING | — |
| negative_prompt | STRING | — |