Chinese phone manufacturer Honor has launched an image-to-video AI generator powered by Google, before it’s available to Gemini users. It will be available first for anyone who buys the Honor 400 or 400 Pro phones, which launch next week on May 22nd.

The new AI tool, powered by Google’s Veo 2 model, creates five-second videos based on static images, in either portrait or landscape, and takes a minute or two to generate each time. The feature is built directly into the Gallery app on the new Honor phones, and is designed to be simple: there’s no option to include a text prompt along with the image, so you’re stuck hoping that the AI does something sensible with it.

Sometimes it works well. Give it a simple subject, like a clear photo of a person or pet, and it can generate quite realistic movement — albeit I’m pretty sure my cat Noodle’s tongue isn’t quite that big. Other subjects prove trickier: faced with a vintage car it made it rotate impossibly on the spot; fresh tomatoes were fondled by a ghostly hand; and it imagined a women’s soccer game with at least 27 players across three teams, with two referees to keep control of the chaos. The first time I tried it, on a self-portrait by Vincent Van Gogh, it decided that the most appropriate thing would be for a pigeon to fly out of his eye.