ComfyUI Node
Gemini Video Omni (Direct API)
A ComfyUI node in Gemini Direct with 10 inputs and 5 outputs.
Gemini Video Omni (Direct API)
- images
- video
- video
- text
- interaction_id
- cost_info
- cache_key
◄prompt►
◄modelgemini-omni-flash-preview►
◄aspect_ratio16:9►
◄seed0►
◄interaction_id►
◄storetrue►
◄api_key►
◄timeout_sec480►
CategoryGemini Direct
Inputs (10)
| Name | Type | Default | Description |
|---|---|---|---|
| prompt | STRING | Describe the scene, dialogue and sound. State the desired DURATION explicitly (3-10s) -- there is no duration widget. Output is fixed at 720p/24fps with audio. | |
| model | COMBO | gemini-omni-flash-preview | Gemini Omni video model. |
| aspect_ratio | COMBO | 16:9 | Output aspect ratio. |
| seed | INT | 00–18446744073709550000 | Re-run trigger only -- NOT sent to the API. Gemini Omni video is non-deterministic regardless of seed; change this to force ComfyUI to re-execute the node. |
| imagesopt | IMAGE | Reference images (batched), up to 14. | |
| videoopt | VIDEO | Reference video to edit (<=10s, >=3s recommended). EEA/CH/UK: editing an UPLOADED video is region-blocked. Wire interaction_id instead to edit a GENERATED video. | |
| interaction_idopt | STRING | Interaction ID of a previous turn to continue/edit server-side. Wire from another Gemini Video Omni node's interaction_id output, or paste one to resume across sessions (expires ~55 days; requires store=True). | |
| storeopt | BOOLEAN | true | Retain this interaction server-side so it can be continued via interaction_id. Required for chaining. Set False for privacy -- breaks chaining on this turn. |
| api_keyopt | STRING | Google AI API key. Falls back to GEMINI_API_KEY env var or gemini_api_key.txt in ComfyUI root. | |
| timeout_secopt | INT | 48010–1800 | HTTP timeout (seconds) for the Gemini API call. Default 480s. |
Outputs (5)
| Name | Type | Description |
|---|---|---|
| video | VIDEO | — |
| text | STRING | — |
| interaction_id | STRING | — |
| cost_info | STRING | — |
| cache_key | STRING | — |