RumpelstiltskinAI.co

Rumpelstiltskin AI prompt for Seedance or Kling

Last updated: 2026-10-09

Use a reference-video prompt that preserves the tiptoe choreography and replaces only the little man with your consenting adult reference. Seedance and Kling controls differ by provider, so attach the reference video and photo using the interface’s actual input fields. A prompt alone does not guarantee an identical clip.

Preview my first frame — free ↗

Copyable reference prompt

Edit @Video1 using @Image1 as the replacement identity. Target the small standing man in the patterned black tailcoat, white shirt and black bow tie: he approaches the seated woman, raises an index finger to his lips, spreads his arms while grinning, and tiptoes on the straw. These original traits identify the target only; do not retain his original face or pale curly hair. Replace his identity in every wide shot and close-up with the same face, hair and facial hair from @Image1. If @Image1 shows the full body, preserve its recognizable body and visible legs while following the source choreography. Keep the seated, long-haired woman in the pale dress unchanged. @Video1 is the complete scene reference: preserve its straw-filled barn, candlelight, shot order, cuts, framing, camera movement, gestures and expression timing. Match the replacement to the original soft film texture and lighting. No extra people, text, logos or watermarks. Keep the source audio as a separate track and restore it after generating the video.

How to use the prompt

First, obtain a reference video and portrait you are allowed to process. Add the motion reference to the model’s video-reference field and the portrait to its character or image-reference field. Paste the text above into the instruction field. The @Video1 and @Image1 names match fal’s Seedance interface; replace those names with the exact reference tokens used by another provider or Kling interface. Some products expose a start frame instead of separate character references; in that case, edit a still first and use it as the image input. Do not paste invented file tokens into a prompt and expect the service to find a local file. A URL or attachment must reach the provider through a supported input. Save the original portrait and reference separately so you can repeat the test.

Keep identity separate from wardrobe

The portrait establishes who the dancer is, while the video supplies the costume, scale, movement and room. Mixing those jobs can lead to a person in their everyday shirt dancing in a changed background. Describe the features that should follow the portrait, such as hair, facial hair and face shape, then explicitly preserve the costume and camera layout from the reference. Avoid demanding a smile or open mouth in every shot; the reference contains a shushing gesture and changing expressions. Copying those expressions is part of the recognizable sequence, not an error to remove.

Why a single whole-clip instruction is difficult

The 13-second reference alternates wide views, close-ups, reaction shots and a foot-only view. A model may retain a face in one view and redesign it in another. It may also redraw the maiden even when you wanted only the little man changed. If a face is obscured by a hand, or appears only from behind, demanding a fully visible face can change the source composition. Review the actual frames rather than using a blanket instruction that both faces must always be visible. That condition does not describe this reference and can cause unnecessary generation failures.

Lock the target before generating the whole clip

Describe the standing dancer by costume, position and recognizable actions, then identify the seated woman separately. The black tailcoat and bow tie locate the source character; they must not instruct the model to preserve the source face or pale curls. State that the reference image is the only identity source for that target. For two-person mode, add a second image and assign it explicitly to the seated, long-haired woman in a pale dress who covers her mouth. Keep these role assignments consistent in every shot. A full-body image can show the intended legs and body, but a prompt cannot guarantee that feet or anatomy will transfer correctly. Inspect those shots in the moving result.

Check the moving result before sharing

A convincing first frame does not prove a convincing video. Watch the eyes, mouth, fingers, face size and transitions at normal speed and frame by frame. If a replacement disappears at a cut, revise the identity and role instructions before spending on another attempt. Keep the model’s native output frame rate; do not pad it to a higher rate with duplicated frames. Restore the reference audio after generation, trimming the audio if the new video is shorter instead of stretching the image. Compare final duration and check sound alignment with the source. Export the full wide version before cropping; a 9:16 center crop can remove a supporting person, which may be acceptable for solo mode but not for every two-person shot.

Use the one-photo workflow instead

Select your photo for a free first-frame preview. AI preview is being connected; selecting a photo currently shows it locally on your device. RumpelstiltskinAI.co centralizes the complete reference video, explicit character assignments, identity instructions and delivery settings so you do not have to repeat those steps in separate tools. Solo mode takes one portrait for the man; pet mode uses an animal reference, while two-person mode adds a maiden portrait. Default one-time packs are 1 video for $6.99, 3 videos for $14.99, 10 videos for $39.99. Delivery time is being measured with the reference workflow. No turnaround is guaranteed before the generation test passes. Check the generator’s current availability and free-preview explanation before expecting a paid moving result.

Sources and verification

fal Seedance 2.0 Fast reference-video and image input documentation

Checked 2026-10-09. External prices and product rules can change.

Frequently asked questions

What is Rumpelstiltskin AI?+

Rumpelstiltskin AI recreates the tiptoe meme with your own character. RumpelstiltskinAI.co uses one photo in solo mode, optional two-photo mode, and pet mode. The complete generation workflow is being tested.

Does the video include the original sound?+

The intended output preserves the supplied reference clip’s audio. The file contains an AAC track, but its exact recording and reuse license have not been verified. The viral meme is commonly paired with Riff Raff’s “Tip Toe Wing In My Jawwdinz”; that does not identify every uploaded clip or grant music rights. See our song guide.

How long does a video take?+

Delivery time is being measured with the reference workflow. No turnaround is guaranteed before the generation test passes.

Can I use it for free?+

The free AI first-frame preview is being connected. You can select a photo for a local preview now, without signing in or paying. Full video generation will be paid.

Can I buy without signing in?+

Email guest checkout is the selected workflow. It will open after generation and payment tests pass; Google and email accounts are available separately.

How much does a video cost?+

The default pack prices are $6.99 for one video, $14.99 for three, and $39.99 for ten. Each standard video uses 10 credits. Your displayed checkout price, including any applicable tax, is authoritative. Checkout is closed until end-to-end testing passes.

Can I post it on TikTok or Reels?+

Yes, the default 9:16 format fits TikTok and Instagram Reels. Follow the platform’s AI-disclosure and music rules, use photos with permission, and check audio rights before posting. A downloadable MP4 does not grant blanket rights to third-party music.

How do pet and two-person modes work?+

Pet mode uses one clear animal photo for the little man’s role. Two-person mode uses separate adult reference photos for the man and maiden. Solo mode uses your photo for the man and a template reference for the maiden. The complete scene is regenerated from the reference video. Pet and two-person output are released after their own quality tests pass.

What happens if generation fails?+

Paid checkout remains closed while automatic credit restoration is being verified. For an undelivered or defective paid result, email ronfenrir@gmail.com with your order identifier for review under our refund policy. Credit restoration and a card refund are different actions.

How long are my photos stored?+

Selected photos currently stay on your device. When AI uploads are enabled, the configured retention is 24 hours for inputs and 30 days for completed videos; the deletion process must pass testing before this becomes an active service promise.

How is this different from other Rumpelstiltskin AI tools?+

RumpelstiltskinAI.co is an independent site and is not affiliated with rumpelstiltskinai.com. Our selected workflow starts with one photo, offers a free first-frame preview without sign-in, uses one-time packs starting at $6.99, defaults to 9:16, and includes pet and optional two-person modes. Availability is shown on the generator; we do not claim every mode has passed testing yet.

Do I need to cancel a subscription?+

No. The advertised packs are one-time purchases with no automatic renewal. For an unused purchase or a refund request, email ronfenrir@gmail.com with your order identifier; the refund policy and mandatory consumer rights apply.

What photo should I use?+

Use a sharp, front-facing photo of one consenting adult or your pet. A clear full-body photo can also guide visible clothing and legs; it does not guarantee perfect body replacement. Avoid side views, heavy filters, sunglasses, blur, explicit material, minors, public figures, and photos you do not have permission to use.

Select your photo for a free first-frame preview. AI preview is being connected; selecting a photo currently shows it locally on your device.