Advance Gemini icon

Advance Gemini

Interact with Google Gemini AI models

Actions4

Generate Video

AI-generated

Summary

Generate a video with a Veo model from a text prompt, frame images, reference images, or an existing Gemini video. The node polls until generation completes, downloads the first generated video, uploads it to S3, and returns its URL and generation details.

Inputs

  • Video Model — Select Veo 3.1 Fast (default) for faster generation or Veo 3.1 for higher quality.
  • Generation Mode — Choose text to video, frames to video, references to video, or extend video.
  • Video Prompt — Text description of the video. Required at runtime for text to video; optional in the other modes. A negative prompt can describe content to exclude.
  • Aspect Ratio — Choose 16:9 or 9:16 for text, frame, and reference modes. The setting is not used when extending a video.
  • Resolution and Duration — Choose 720p or 1080p. At 720p, select 4, 6, or 8 seconds; at 1080p, duration is 8 seconds. The default resolution and duration are 720p and 8 seconds.
  • Person Generation — The mode-specific option is Allow All for text and extension modes, or Allow Adult for frame and reference modes. Region restrictions may apply.
  • Generate Audio — Defaults to enabled. When disabled, the node adds silent-film instructions to a non-empty prompt and attempts to remove audio from the downloaded video using ffmpeg.
  • Frame images — For frames to video, provide a required start frame URL. Optionally provide an end frame URL, or enable looping to use the start image as the end frame.
  • Reference images — For references to video, provide at least one reference image URL. An optional style image URL can also be supplied.
  • Input Video URI — For extend video, provide the required Gemini video URI as a full URL, files path, or file ID.
  • S3 Bucket Name (required) — Bucket to which the generated MP4 is uploaded. S3 credentials are also needed for the upload.
  • S3 Public Domain — Optional domain used to construct the returned video URL instead of the default S3 URL.
  • Additional Options — Set the polling interval (default 10 seconds, range 5–60) and maximum wait time (default 30 minutes, range 5–60).

Output shape

One n8n item per input item, with JSON metadata and no binary video field.

The video is uploaded to S3; the JSON includes videoUrl, fileName, model, resolution, generationMode, and Gemini's video URI, plus applicable prompt, aspect ratio, duration, and person-generation details. Only the first generated video is downloaded and uploaded. If audio generation is disabled, the output reports whether ffmpeg removed the audio; if that attempt fails, the video may still contain audio.

Links

Discussion