ComfyUI Node
Video Prompt Dexogen External
Writes a target-aware video prompt with an external OpenAI-compatible language or vision model. With no image connected it writes from the idea. One image becomes the first-frame reference; two images become the first and last anchors. The target profile supplies mode-specific formulas, duration guidance, and the frame grid used for the frames and seconds outputs. The API key is optional and stored outside the workflow behind an opaque per-node reference. Direct prompt mode bypasses the endpoint and image inputs, and passes the entered prompt through exactly as written while retaining the target frame-grid outputs.
Video Prompt Dexogen External
- first_frame
- last_frame
- text
- frames
- seconds
Category🧬 Dexogen/💬 Prompt & Text
Inputs (2)
| Name | Type | Default | Description |
|---|---|---|---|
| first_frameopt | IMAGE | The starting picture. Connecting one image switches the node to first-frame mode. | |
| last_frameopt | IMAGE | The ending picture. With both images connected, the node writes the transition from the first to the last. |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| text | STRING | The generated prompt as a plain STRING. In Direct prompt mode, or without an endpoint or model, the entered text passes through unchanged. |
| frames | INT | The requested duration converted to the target profile's valid frame grid. |
| seconds | FLOAT | The real duration after frame-grid snapping, calculated as frames / fps. |