How to make a Rumpelstiltskin AI video

Step one: pick your pair. Photo 1 becomes the little man - the performer who tiptoes in, shushes the camera and dances. Photo 2 becomes the maiden, who watches from the hay and reacts. Both are real roles, so a good pair has energy: a friend who will commit to the joke, a pet, a partner, or someone who would be funny as a tiny dancing stranger. Pick the photos before you buy credits - a bad photo wastes a render.

Step two and three: upload, choose, generate. Each photo slot takes one subject: face toward the camera, no sunglasses or hats, ideally head to the waist so the outfit comes through, and no mirror selfies (the phone tends to appear in the render). Our pre-check runs the moment both photos are in and tells you if one cannot be used, before anything is charged. Then pick Standard (50 credits, 480p) or HD (100 credits, 720p), press generate, and wait about 30 seconds to two minutes. The result lands in your library as a silent 16:9 MP4 - add the trending sound in TikTok or Reels, label it as AI, and post.

If a render disappoints, work through four checks in order: is each photo one clear front-facing subject (re-crop closer if the face is small), is the lighting even rather than backlit or heavily filtered, does each photo show the outfit you want, and are you looking at a 480p render that will simply look softer than the HD version. Photo quality, not settings, is behind almost every disappointing result - which is why the pre-check runs before anything is charged.