How our preset scene works without prompts
On AI Rumpelstiltskin, the straw-filled room, tiptoe choreography, and camera angles are already configured. There is no prompt box to fill out and no option to upload a custom reference video. As shown in our creation guide, you sign in, check generation is available before paying, pick your role mode, and upload one photo for a single character or two separate photos when you Choose Both. Once you select a 6-second or 13-second duration and an available frame ratio, the tool generates the video and adds the soundtrack automatically.
Why other workflows discuss prompts and reference stills
General-purpose video tools often require extra setup steps that do not apply here. For instance, Media.io's tiptoe generator page outlines workflows that combine one or two photos with a compatible reference video and text prompts to distinguish building a new scene from replacing characters in an existing clip, while noting that naming a song in a prompt does not produce the original audio recording.
Similarly, in a Reddit thread on character-swap videos, a creator trying to replace both people across multiple cuts asked whether to supply a raw portrait or an already-edited still frame after running into missed shots and rising costs. On our site, you only provide clean single-person portraits—see our photo upload requirements for file tips, and read why faces can change across cuts to understand multi-shot consistency.
Unverified sample prompt for external generators
If you are experimenting with another generator that accepts image references alongside text instructions, you can adapt the original example below. It has not been tested on third-party models and will not guarantee matching choreography, likeness, or music:
"Inside a dimly lit wooden room with straw scattered across the floor, Subject A performs a sneaky exaggerated tiptoe dance across the center while Subject B sits still on a wooden bench watching calmly, warm indoor lighting, cinematic medium-full shot."