ComfyUI Node
PromptForge Enhancer
A ComfyUI node in đź”® PromptForge with 19 inputs and 2 outputs.
PromptForge Enhancer
- model
- image
- image2
- image3
- video
- ENHANCED_PROMPT
- THINKING
â—„prompt_textâ–ş
â—„modetext_onlyâ–ş
â—„enhancement_presetautoâ–ş
â—„seed1â–ş
â—„max_tokens2048â–ş
â—„temperature0.70â–ş
â—„top_p0.90â–ş
â—„top_k0â–ş
â—„num_beams1â–ş
â—„repetition_penalty1.10â–ş
â—„system_prompt_overrideâ–ş
â—„video_seconds5â–ş
â—„frame_count16â–ş
â—„negative_promptâ–ş
Categoryđź”® PromptForge
Inputs (19)
| Name | Type | Default | Description |
|---|---|---|---|
| model | PROMPTFORGE_MODEL | MODEL output from a PromptForge Model node (source, model, quant and runtime settings). | |
| prompt_text | STRING | Raw user request / media-derived text to enhance. Can be loose natural language, random idea, detailed description, keyword pile, anything. | |
| mode | COMBO | text_only | text_only ignores all media; inputs uses every connected image (image / image2 / image3) and the sampled video frames. |
| enhancement_preset | COMBO | auto | Preset system prompt: auto, or one of the PromptForge presets (Krea 2 T2I, MOSS Image T2I, Qwen Image T2I, generic styles...). |
| seed | INT | 11–4294967295 | Sampling seed. |
| max_tokens | INT | 204832–65536 | Max new tokens for the enhanced prompt. Keep it below the model context for GGUF. |
| temperature | FLOAT | 0.700.1–1 | Sampling temperature. |
| top_p | FLOAT | 0.900–1 | Nucleus sampling cutoff. |
| top_k | INT | 00–100 | Top-K. 0 = model default. |
| num_beams | INT | 11–8 | Beam search width. 1 = fast sampling; >1 trades speed for stability. |
| repetition_penalty | FLOAT | 1.100.5–2 | Repetition penalty. |
| system_prompt_override | STRING | Custom system prompt override. Leave empty to use the selected preset. | |
| video_secondsopt | INT | 51–60 | Target video duration in seconds. |
| imageopt | IMAGE | First reference image. | |
| image2opt | IMAGE | Second reference image / edit source. | |
| image3opt | IMAGE | Third reference image / edit source. | |
| videoopt | IMAGE | Video frames input. Use frame_count to control how many frames are sampled. | |
| frame_countopt | INT | 161–64 | Frames sampled from video input. |
| negative_promptopt | STRING | What the generated prompt/scene must not contain. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| ENHANCED_PROMPT | STRING | — |
| THINKING | STRING | — |