Decide whether you are inventing or editing the scene
A text-to-video prompt asks a model to construct a window scene. A reference edit already has a building, dancer and camera path; its main task is to change the person consistently. The EasyVideoAI template follows this second workflow.
For a reference edit, avoid requesting a different room, extra dancers or a new camera cut. Those instructions work against the supplied performance. If you want to invent a different scene in another tool, its reference controls and output will differ from this fixed template.
An original reference-edit prompt example
Replace only the dancer behind the illuminated apartment window with the person in the uploaded photo. Keep the photo's recognizable face, hair and outfit on that same dancer from the distant opening through the closer view and turn. Preserve the dark building, warm room, window frame, small glowing speaker, camera approach and original dance timing. Keep the smile toward the lens before the curtain pull, and preserve hand contact as the curtain covers the window. Do not introduce people, camera cuts, captions or a different ending.
This example explains the intended edit; it has not been separately quality-tested as a standalone prompt. Use it only in a tool that supports a video reference and portrait. EasyVideoAI has no prompt box for this template because the role and scene instructions are already included.
Keep the same identity as the camera gets closer
This scene makes identity continuity harder to judge at the start: the person initially occupies a small part of the frame. Review the face when the window fills more of the picture and when the dancer looks toward the lens.
Choose a portrait with visible hair and clothing as well as the face. A face-only crop leaves the outfit less specified. The uploaded background is not transferred; the prepared room remains the setting. Asking the prompt to copy both locations can introduce conflicting directions.
Treat the curtain as part of the performance
The curtain is the ending, not background decoration. Its movement should follow the dancer's hand, cover the opening at the right point and preserve the original sequence. A curtain that closes during the first dance movement loses the scene's reveal.
Check the window frame, hand and fabric together. Prompts can guide their relationships, but cannot guarantee flawless contact or anatomy. If a completed result is unsatisfactory, inspect what went wrong before deciding whether a different photo or another paid generation is appropriate.
Separate a TikTok prompt from soundtrack selection
The phrase Window Dance AI prompt for TikTok tells you the intended destination, not which song will appear. This template restores audio from the prepared clip. It does not choose a live trending sound based on the text prompt.
The output keeps a vertical composition. If you add text or other audio after downloading, review the exported file again so captions do not hide the reaction and the new track does not accidentally double the supplied sound.
Reviewed against the EasyVideoAI studio on November 5, 2026. Check the live studio for current availability and the final credit quote.
