Turn one photo into a steady video greeting with Wan 3.0

A still camera and small motion: Wan's home ground

Wan is Alibaba's family of video models, and the version we run is Wan 3.0 Prime. It takes one start picture or a written description and returns a clip of 5 or 10 seconds at 720p or 1080p. Our catalogue tags it "Steady loops", and that is the honest summary: it works best when the camera holds still and the motion is small and even.

That suits more greetings than you might expect. An old couple on the bench where they met, leaves coming down around them. A newborn asleep under a slowly turning mobile. Snow drifting past the window while the family lifts their cups. People watch clips like these twice, and a steady frame keeps the eye on the faces.

It is not the model for a burst of confetti, a crowd jumping to its feet or a line spoken to camera. For bigger, busier motion, Kling 3.0 is the better pick. For a family singing Happy Birthday, pick Veo 3.1: it can make the singing along with the picture. For someone saying your greeting with their own face, make a talking card.

A quiet ten seconds or a busy five: Wan against MiniMax, Kling and Veo

All four turn one of your photos into a video, and all four can give it sound. They differ in how the clip feels, how long it runs and what it costs per second. MiniMax H3 is our Auto choice and the best value here: it always adds sound, animates every video scene in the catalogue and runs from 5 to 10 seconds. If you have no strong preference, start there.

Pick Wan 3.0 Prime when the moment is quiet and you want the frame to stay calm, or when you want ten full seconds, which Veo 3.1 cannot give. Pick Kling 3.0 when more is going on - glasses clinking along a garden table, friends throwing confetti at a picnic - or when you need up to 15 seconds. Pick Veo 3.1 for the cinema look and for voices and singing made together with the picture; with sound on it costs more per second, and it stops at 8 seconds.

Brief Wan like a patient camera operator

Wan animates what is already in the frame, so the start picture does half the work. Choose a photo where the faces are well lit and not too small. If the people you want in the clip live in different cities, make a photo card first - up to six people in one picture, with faces most reliable up to four - then bring that card to life with Wan.

Then write the clip the way you would brief a patient camera operator. Say who moves and how much: "he takes her hand" works, "they are happy" does not. Say what the camera does; "steady camera" or "the camera does not move" plays to Wan's strength. If sound is on, name one or two sounds, not a whole soundtrack. Use five seconds for a single gesture and ten for a slow dance or a long look out to sea. One clear movement looks more natural than five small ones competing for the same few seconds.