In addition to images and audio, Google is now incorporating video generation into the Gemini app, utilizing its Veo 2 model for Advanced subscribers.
Introduced at the close of last year, Veo 2 boasts “fluid character movement, lifelike scenes, and enhanced visual details across various subjects and styles,” along with “cinematic realism,” achieved through a deep understanding of real-world physics and human motion.
Within Gemini, Veo 2 is capable of generating eight-second video clips at a 720p resolution. You will receive an MP4 file downloadable in a 16:9 landscape format. Additionally, you can share your creations via a g.co/gemini/share/ link. To input your prompt, select Veo 2 from the model dropdown menu available in both web and mobile applications.
To create your desired scene, provide a description: “The more detailed your description, the greater control you have over the resulting video.” The clip will take approximately 1-2 minutes to generate.
This presents a realm of exciting creative prospects, allowing your imagination to run wild with unique combinations, exploring a range of visual styles from realism to fantasy, or quickly narrating concise visual concepts.
Here are some example prompts and the outputs we created:
- “small puppy running through a garden filled with thick snow in the early morning sun”
- “a kitten being introduced to the beach, specifically near the water, and then getting surprised as a wave crashes nearby”
- “Aerial shot of a grassy cliff overlooking a sandy beach where waves crash against the shore, featuring a notable sea stack rising from the ocean close to the beach, illuminated by the warm golden rays of either sunrise or sunset, showcasing the dramatic elevation change and tranquil beauty of the Pacific coastline.”
Regarding safety, each frame includes a SynthID digital watermark.
This feature is exclusively for Gemini Advanced subscribers at $19.99 per month. There’s a “monthly limit” on how many videos users can produce, with notifications being sent when they approach that limit. This capability is rolling out globally across all supported languages for Gemini, starting today, with full availability expected in the following weeks.
In parallel, Google One AI Premium subscribers now have access to Veo 2 generation within Whisk. Announced in December, this experiment from Google Labs enables users to “prompt with images” instead of text. The recent update introduces “Whisk Animate,” allowing the images you create to be transformed “into vivid eight-second videos using Veo 2.”
