Rumpelstiltskin AI prompt (copy and paste)
A free Rumpelstiltskin AI prompt to copy: a timed shot list for the creep, shush, grin and tiptoe dance, plus a two-part version, variants and a fix-it table.
8 最小読み取り · 更新済み
The prompt
Paste this into an AI video tool with an image-to-video or reference-image mode, and attach a clear photo of your face as the reference. It’s written in our own words from the meme’s shots, free to use. Rather skip the prompt? The Rumpelstiltskin page makes the scene from one photo.
Rumpelstiltskin AI prompt (15 s, 16:9)
Widescreen 16:9, 15 seconds, photoreal, looking like a grainy 1980s fairy-tale film: warm faded color, soft focus, film grain. Night inside an old wooden barn heaped with golden hay, lit by one lantern. The camera never moves; the edit uses quick hard cuts. The dancer has the face and hair of the person in my reference photo, in every shot, including close-ups. The dancer is small and crouched, wearing a black velvet tailcoat with long tails, a white shirt, a black bow tie, black knee breeches and black shoes whose long pointed toes curl upward. A young woman with long blonde hair in a white nightgown sits in the hay, mostly seen from behind. She is a different person. 0–1 s: from behind the woman’s head, the dancer sneaks toward her through the hay on tiptoe, a finger lifted. 1–2 s: close-up, the dancer presses a finger to their lips and gives the camera a sly look. 2–3 s: the woman clasps a hand over her mouth, eyes wet. 3–4 s: a very close, wide-lens shot of the dancer’s face splitting into a huge, eerie grin. 4–6 s: wide shot, the dancer flings both arms out and bounces from side to side on tiptoe, tails flapping. 6–15 s: cut back and forth between a close-up of the dancer swaying with arms out and still grinning, and a low shot of the curled shoes pattering through the straw. Sound: no music or dialogue; I will add the sound later. Avoid: daylight, modern clothes, my own outfit, extra people, a second dancer, warped hands or feet, flicker, text, logos, watermarks.
The prompt can’t add Riff Raff’s song or time the cuts to it. Add the Rumpelstiltskin sound when you post; the sound guide shows how.
Two-part version for tools with short clips
If your tool’s clips are shorter than 15 seconds, make the story half and the dance half separately, with the same photo and the same first two paragraphs of the prompt above, then join them in an editor.
Part 1: the creep, the shush and the grin
Widescreen 16:9, photoreal 1980s fairy-tale film look, grain, warm faded color. Night barn full of golden hay, one lantern, locked-off camera, hard cuts. The dancer has the face and hair of my reference photo in every shot; small, crouched, black velvet tailcoat, white shirt, bow tie, curled pointed shoes. A blonde woman in a white nightgown sits in the hay, seen from behind. Shot 1: from behind her head, the dancer sneaks toward her on tiptoe, a finger lifted. Shot 2: close-up, finger pressed to the dancer’s lips, a sly look at the camera. Shot 3: the woman clasps a hand over her mouth. Shot 4: a very close wide-lens shot of the dancer’s huge, eerie grin. No music. Avoid: daylight, modern clothes, extra people, text, watermarks.
Part 2: the tiptoe dance
Same barn, same lantern, same film look and the same dancer with the face and hair of my reference photo, in the black velvet tailcoat and curled pointed shoes. Shot 1: wide shot, the dancer flings both arms out and bounces from side to side on tiptoe, tails flapping. Shot 2: close-up, arms still out, swaying side to side with a knowing grin. Shot 3: low shot of the curled shoes pattering fast through the straw. Repeat shots 2 and 3, cutting on a steady beat. No music. Avoid: extra feet, melting shoes, warped hands, text, watermarks.
How to use it
- Pick a tool with an image-to-video or reference-image mode: Kling, Higgsfield, Runway, Veo or CapCut’s AI video all have one in some form.
- Attach one clear photo of your face. Don’t describe your face in words; the photo does that better.
- Paste the prompt and set 16:9 and the longest length you have.
- Generate, watch the close-ups first, and use the fix-it table below if something is off.
- Download and post with the Rumpelstiltskin sound.
What each line of the prompt is for
Each timed line matches one shot of the meme. Here’s what they look like when they come out right:
- The “one lantern, night” line keeps the warm, dark film look. Without it, tools drift to a bright, clean barn.
- “The camera never moves” matters: the original is locked off, and a moving camera makes it look like a music video.
- The costume is spelled out item by item because tools otherwise copy the clothes in your photo.
- “She is a different person” stops the tool from giving the woman your face.
Variants
Keep your glasses or beard
Add to the second paragraph
Keep the glasses and beard from my reference photo in every shot, including the shush and grin close-ups.
Two photos: you as the dancer, a friend as the girl
Replace the third paragraph
The young woman in the hay has the face and hair of the person in my second reference photo, in a long white nightgown. She and the dancer are two different people; never swap or merge their faces.
Keep your own clothes
Replace the costume sentence
The dancer wears exactly the outfit from my reference photo, but keeps the curled pointed shoes.
The own-clothes version is funnier for some people (a hoodie in a fairy-tale barn), but the velvet tailcoat is what makes it read as the trend at a glance.
Fix-it table
| Symptom | Change to the prompt |
|---|---|
| Face drifts in the close-ups | Add: “same face as my reference photo in every close-up” |
| Bright, modern-looking barn | Add: “night, one lantern, deep shadows, 1980s film grain” |
| Dancer wears your clothes | Add: “not the clothes from my reference photo” |
| The woman has your face | Add: “she is a different person, never the dancer” |
| Feet melt or multiply | Shorten the shoe shots; add “two feet, clean steps” |
| One long take, no cuts | Add: “hard cuts between every shot”, or use the two-part version |
| Blocked or refused | Use a plainer photo; remove words like “creepy” or “scary” |
Photo tips for the prompt route
- One face, looking at the camera, in soft light. The close-ups magnify every flaw.
- Eyes visible: the shush and the grin are all eyes.
- No sunglasses, masks or heavy filters.
- Only your own photo or one from someone who agreed; never a celebrity or anyone under 18.
The photo guide has do and avoid examples.
What a prompt can’t do
- Add the song. AI video tools make their own sound or none; Riff Raff’s track comes from the app when you post.
- Hit the exact cuts of the meme. The real edit loops the sway and the shoes on the beat; a generated clip only gets close.
- Make it the same twice. Every run is different, so keep the take you like.
The no-prompt route
If you’d rather skip the retries, the Rumpelstiltskin page makes the whole 15-second scene from one photo, with every cut in place, your face on the dancer and the trend’s sound laid over it, in about 5 minutes. No watermark.
One Standard video uses 90 credits. The smallest pack that makes one is $12.90 for 90 credits (1 video), a one-time buy whose credits never expire, and plans start at $19.90 a month for 270 credits. If a video fails, its credits come back. Compare the routes in CapCut vs AI vs one photo. NowStarring is an independent fan tool, not linked to the creators of the original film or to Riff Raff. The song belongs to its rights holders.
よくある質問
What is the prompt for the Rumpelstiltskin AI trend?
A timed 16:9 shot list: the dancer creeps in over the girl’s shoulder, shushes the camera, grins in a wide close-up, dances on tiptoe with arms out, then the sway and the curled shoes in the straw cut back and forth. Attach your photo as the reference instead of describing your face.
Which AI tool should I use for the Rumpelstiltskin prompt?
Any tool with an image-to-video or reference-image mode, such as Kling, Higgsfield, Runway, Veo or CapCut’s AI video. Pick one that keeps faces steady in close-ups.
Why doesn’t my Rumpelstiltskin AI video look like me?
Usually the photo: a small, dark or filtered face loses detail in the close-ups. Use a sharp, front-facing selfie in soft light and add a line asking for the same face in every close-up.
Does the Rumpelstiltskin prompt include the song?
No. AI video tools don’t add Riff Raff’s track. Pick the Rumpelstiltskin sound in TikTok or Reels when you post.
Can I make the Rumpelstiltskin video with two people?
Yes, with the two-photo variant: you as the dancer and a friend as the girl in the hay. Tell the tool they’re two different people so it doesn’t merge the faces.
Is the prompt the one used for the original video?
No. The creators of the original never shared their tool or prompt. This is our own shot list written from the meme.
