SelfScene turns 1–4 of your own photos and a short scene description into a brand-new 5–15 second AI video with sound. It never edits existing footage and never swaps faces or replaces people in a video. Image to Video generates the clip with the MiniMax H3 reference-to-video model through fal; Video to Prompt can describe a clip you like (shot, camera, composition, light and sound) as a prompt. fal lists the generation model at $0.05 per second at 480p and $0.06 at 768p; here a second costs 10 or 12 credits.
Want the details? Read the photo-to-video guide or how photo to video works.
