HiAPI
Showcases

Reference Image-Driven Short Video: Gemini Omni Flash Parameter Illustration

A demonstration of short video generation using Gemini Omni Flash's reference-to-video pipeline to show how "multiple reference images control the subject, scene, and style." The example submits 3 reference images that respectively lock in the character subject, the scene environment, and the overall color tone, and explains the role of each image in the prompt. Combined with output parameters of 8-second duration, 720p, and 16:9, it presents a continuous camera-movement shot of city life. It is suitable for developers to understand how to arrange reference images in the image_urls array, how to choose duration and resolution parameters, and how the number of reference images affects visual consistency; it can also serve as a starting template for short video assets and product demo clips.

Style reference-to-video · multi-shot · Technical Explanation

Model Gemini Omni

Reference inputs

  • image_urls

    image_urls

  • image_urls

    image_urls

  • image_urls

    image_urls

Prompt

Try this model
A single continuous long take: a young woman walks through a modern city street during golden hour, the camera tracks alongside her at eye level with a slow dolly follow, shallow depth of field turns the passing lights into soft bokeh, warm amber and teal color grading, subtle handheld micro-movement, the subject's identity and clothing remain consistent throughout, the environment and atmosphere match the reference scene, cinematic 16:9 composition, 8 seconds, smooth and fluid motion, no cuts.

Copy the full prompt and adapt it for your own generation.

Tags