ComfyUI Node
TS Super Prompt
Voice prompt field with optional Qwen3.5 AI prompt enhancement for image, video, and music prompts. Attach an optional reference image right in the node (drag-drop / paste / file picker) — it is used by the Enhance button without running the workflow.
TS Super Prompt
- images
- text
◄text►
◄high_qualityfalse►
◄system_presetPrompts enhance►
◄attached_image►
◄attached_image_2►
◄bigger_modelfalse►
CategoryTS/LLM
Inputs (7)
| Name | Type | Default | Description |
|---|---|---|---|
| text | STRING | Prompt field: recognized speech lands here, and the Ai prompt button replaces the text with an enhanced prompt. | |
| high_quality | BOOLEAN | false | Enable to transcribe speech with Whisper turbo (large-v3 turbo). Off: uses the fast base model. |
| system_preset | COMBO | Prompts enhance | System preset from qwen_3_vl_presets.json used to enhance the prompt. |
| attached_image | STRING | Internal field: annotated path of the attached image (filled by the Attach button in the node). | |
| attached_image_2 | STRING | Internal field: annotated path of the second attached image. With two images the first is the FIRST frame and this one the LAST frame of the shot. | |
| bigger_model | BOOLEAN | false | Off: the 2B prompt model (fast, already downloaded). On: the 4B one — better prompts, about twice the download and twice the VRAM. The 4B model is fetched on first use. |
| imagesopt | IMAGE | Optional reference images from the graph. A single image is a plain reference; in a batch the FIRST image is the first frame of the shot and the LAST is the last frame. Takes precedence over images attached in the node. Each frame is shrunk to about 1 MP on the way in. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| text | STRING | Prompt text (enhanced when enhancement runs). |