The person from the photo takes the place of the person in your video: the same place, light, camera and movements, with the face and hair from the photo. Photo + video; the clip is 5 s, framed to the video's aspect. Works best when the person in the video is in a medium shot.
Upload a photo: JPG, PNG or WebP, straight from your phone.
The neural network makes the clip, usually in a few minutes.