A prompt cannot add a dance that the source scene does not contain. What it can do is protect the parts of the photo you care about: the face, the outfit, the glasses, the makeup. It can also tell the model what to avoid: extra people, swapped identities, text, watermarks.
The prompts below are deliberately specific. "Make it cool" gives the model nothing. "Preserve the face, keep the black jacket, do not add a second person" gives it three constraints it can actually check. Use one prompt at a time and change one variable between runs.
How to use a prompt without wasting credits
Pick the prompt that matches the thing you are most afraid of losing. If the face matters most, use the identity lock. If the costume matters most, use the outfit lock. Do not combine five prompts into one paragraph; the model will follow the strongest constraint and ignore the rest.
Change one variable between runs: the photo, the scene or the prompt. If you change all three, you cannot tell what caused the improvement. Keep a note of the scene and prompt for the clip you like, so you can reproduce it.
Why prompts cannot invent choreography
The tool edits a prepared performance. The dance comes from the source scene, not from the sentence you type. A prompt that says "do the running man" will not add a running man to a scene that does not contain one. It can only preserve, constrain and style what is already there.
That is a feature, not a bug. It is why the motion is coherent and the camera is stable. If you need a specific move, choose a scene that already contains it, or film yourself and use a motion-control workflow.
Safety and consent
Only upload photos of adults who have agreed to the use. Do not use a photo of a public figure, a coworker or an ex-partner without permission. The consent checkbox is not a formality; it is the line between a creative tool and harassment.
Do not use prompts to remove clothing, change someone’s body or create sexual content. The input and output moderation will block it, and the account can be suspended. Keep the clip something you would be comfortable showing the person in the photo.
01
Clean identity lock
Any solo scene where the face must stay recognisable.
Preserve the uploaded person’s face, facial proportions, skin tone and hairstyle exactly. Keep the same number of people as the photo. Do not add a second person, do not replace the face with a different identity. Keep the scene’s original camera, lighting and body movement. No text, no logos, no watermark.Use this prompt in the studio
02
Whole outfit lock
A costume, wedding dress or streetwear look that must survive.
Preserve the uploaded person’s face and the full outfit, including colour, fabric and visible accessories. Replace the performer in the scene with this person. Keep the scene’s camera and choreography. Do not invent new clothing, do not change the shoes, do not add jewellery that is not in the photo.Use this prompt in the studio
03
Halloween makeup lock
Face paint, prosthetics or a makeup-led costume.
Preserve the uploaded person’s face and the Halloween makeup exactly, including colour placement and shape. Keep the hair and any visible costume pieces. Place the person in the scene without changing the makeup. Avoid smoothing the skin so much that the makeup disappears. No extra people, no text.Use this prompt in the studio
04
Couple identity separation
A two-person duet with two separate photos.
Use the first uploaded photo for the performer on the left and the second photo for the performer on the right. Preserve each person’s face and hairstyle. Do not swap the two identities at any point. Keep the scene’s interaction and camera. Do not add a third person.Use this prompt in the studio
05
Shuffle footwork emphasis
A full-body scene with visible shoes.
Preserve the uploaded person’s face, outfit and shoes. Keep the full body in frame and do not crop the feet. Follow the scene’s footwork and bounce. Keep the arms loose and natural. Do not turn the clip into a close-up portrait. No text, no logos.Use this prompt in the studio
06
Hardstyle silhouette
A wide scene with festival lighting.
Preserve the uploaded person’s face and the silhouette of their outfit. Keep the wide stance and the raised arms from the photo. Use the scene’s high-contrast lighting and camera. Do not soften the look into a beauty portrait. No extra limbs, no text, no watermark.Use this prompt in the studio
07
Glasses and hair
A photo where glasses or a distinctive hairstyle define the person.
Preserve the uploaded person’s face, glasses and hairstyle. Keep the glasses on the face in every frame. Do not change the hair colour or length. Keep the scene’s camera and movement. Avoid warped frames around the glasses. No text, no logos.Use this prompt in the studio
08
Private / no gallery
A personal clip you do not want in the public gallery.
Preserve the uploaded person’s identity and outfit. Keep the scene’s original camera and choreography. Do not add text, logos or watermarks. This result is private and should not be offered to the public gallery.Use this prompt in the studio
09
Costume colour accuracy
A costume whose colour is the joke or the point.
Preserve the uploaded person’s face and the exact costume colours, including small accents. Do not shift the hue or add new colours. Keep the scene’s lighting but do not let it wash out the costume. No extra people, no text, no watermark.Use this prompt in the studio
10
No extra limbs
Any fast arm scene where hands tend to multiply.
Preserve the uploaded person’s face and outfit. Keep exactly two arms and two legs. Do not duplicate hands, fingers or limbs. Follow the scene’s movement. If the hands blur, keep the blur natural rather than inventing extra fingers. No text, no logos.Use this prompt in the studio
11
Vertical social crop
A clip that will be cropped to 9:16 after generation.
Preserve the uploaded person’s face and outfit. Keep the head and hands inside the central vertical safe area so the clip survives a 9:16 crop. Do not add text or UI elements. Keep the scene’s camera and lighting.Use this prompt in the studio
12
Wedding outfit lock
A dress or suit that must stay recognisable.
Preserve the uploaded person’s face and the wedding outfit, including the dress or suit silhouette, colour and visible details. Keep the scene’s camera and dance. Do not change the outfit into a generic costume. No extra people, no text, no watermark.Use this prompt in the studio
Frequently asked questions
Do these prompts change the dance?
No. They constrain identity, outfit, framing and failure modes. The choreography comes from the prepared scene. Choose a different scene if you want different movement.
Can I use these prompts in another AI video tool?
The structure is portable, but the wording is tuned for this photo-to-dance workflow. Other tools may treat the same words as a text-to-video prompt and invent a different scene. Test one prompt at a time.
Where do I paste a prompt?
Use the optional direction field in the studio. Keep it under 500 characters. The prompt is added to the system instruction that protects identity and consent.
Are the prompts free?
Yes. Reading and copying them is free. Generating a video costs 30 credits ($9.90) because it runs a paid model. Failed jobs return their credits.
Can I use the results commercially?
You need the rights to the photo and permission from everyone shown. The prompts do not grant rights to costumes, trademarks or music. Keep commercial use to material you own.
