Rumpelstiltskin AI

How to Make a Rumpelstiltskin AI Video of Yourself

Updated October 13, 2026

What the Rumpelstiltskin AI video is

The clip looks like a lost 1980s fantasy film: a barn full of hay, a blonde maiden in tears, and a tiny imp in a velvet suit who puts a finger to his lips, grins, and tiptoes through the straw with his arms out. The full AI film runs about 33 seconds and ends with gold and a prince; the part everyone copies is the 10-second dance in the middle, cut into 11 shots that land on the music. Since it went viral in late September 2026, people have been putting themselves, their friends and their boss in the imp's place.

So "making a Rumpelstiltskin AI video" usually means one thing: the same scene, the same shots, with you as the dancer.

Can you make it with a prompt?

Not shot for shot. You can get a barn, a maiden and a dancing man from a prompt, but three things don't come from words:

  1. The shot list and timing. The dance has 11 shots in 10 seconds: wide shush, close-up shush, the maiden covering her mouth, the sly grin, the tears, the tiptoe, the eyes, the arms-out dance, the laugh, the feet, and the maiden again. A text-to-video model invents its own camera moves and usually gives you one or two shots.
  2. You, from head to toe. Image-to-video tools animate your photo where it is, so the background stays your bedroom. Face-swap apps paste your face onto the original, so the suit, the legs and the shoes stay someone else's.
  3. Length. Video models that copy a reference clip follow it well for about 5 seconds, then the motion drifts. 10 seconds in one go comes out jumbled.

If you still want to try a prompt in a general video app, this one gives a clip in the same style:

1980s fantasy film still, warm film grain, a wooden barn full of hay and
straw. A crying blonde maiden in a simple dress sits in the hay. A man
in a velvet suit puts a finger to his lips, smiles slyly, then tiptoes and
dances through the hay with his arms out. Soft golden light, handheld
cinematic camera, several cuts between close-ups and wide shots.

Expect a similar mood, not the viral video with you in it.

How we make it shot for shot

Here is what happens after you press Generate on the Rumpelstiltskin AI video generator:

  1. Your photo, completed. If the photo only shows your head and shoulders, an image model first makes a full-body photo of the same person (clothes continued down to trousers and shoes). The video model gets both: your original photo for the face, the full-body one for the rest.
  2. The original clip as a reference, in two halves. The clip is split at a cut 5.5 seconds in, so each half stays close to the length the model can follow. For each half, a reference-to-video model recreates every shot and puts you in place of the imp. The maiden's shots stay as they are.
  3. Back on the beat. The two halves are slowed or sped up to the exact length of the original shots, joined, and put back on the clip's soundtrack. In our tests every cut lands within 0.4 seconds of the original.
  4. Your size. The scene is filmed wide (16:9) like the original. If you picked 9:16 or 1:1, the full shot sits in the middle with a blurred copy of the same frame above and below, so nothing is cropped out.

You get a 10-second video with the original soundtrack, ready to post.

Step by step

  1. Pick a photo. One person, sharp and well lit. Half-body or full-body works best; a front-facing selfie works too. No sunglasses, no group photos, no children or celebrities, and only photos of adults who agreed to it.
  2. Upload it and choose a size. 16:9 for YouTube and X, 9:16 for TikTok, Reels and Shorts, 1:1 for the feed.
  3. Tick the consent box and press Generate. It takes about 5 minutes. You can close the page and find the video in My creations.
  4. Download and post. Captions that work well: "me sneaking to the fridge at 2 a.m.", "when the group chat goes quiet", "me after saying 'one more episode'".

Tips for a better result

  • Show your clothes and shape. The wide dance shots and the close-up of the feet look most like you when the photo shows more than your face.
  • Even light on the face. Hard shadows and hats pulled low make the face drift between shots.
  • Plain is fine. The background of your photo doesn't matter; only the person is used.

Questions

What AI was used to make the original Rumpelstiltskin video?
The creator hasn't said. The look (film grain, warm 80s fantasy lighting, real camera cuts) is typical of current AI video models. You don't need to know which one: what matters for a copy of yourself is following the same shots and timing, which a single prompt can't do.
Is there a Rumpelstiltskin AI prompt I can copy?
There's a short scene prompt in this guide you can try in any text-to-video or image-to-video tool. It gives you a clip in the same style, not the same video: the shots, the cuts on the beat and the dance moves come from the original clip, not from words.
Can I make a Rumpelstiltskin AI video for free?
The guide and the prompt are free. Making the video on this site costs $9.99 (200 credits) (one-time payment, no subscription), and the credits come back if it fails. Free trials on general video apps can get you a similar-looking scene, but not your whole body swapped into the original shots.
Can I make it with my face but keep the original imp's body?
We swap the whole dancer, not just the face, so the hair, clothes and build are yours too. That's why a half-body or full-body photo gives the best result; with a selfie we first fill in the rest of your body.
How long does it take?
About 5 minutes from upload to download. You can close the page; the video appears in My creations.