Burning Bridges AI Prompt
Free to copy. Pick a moment and a length, and paste the result into Kling or any image-to-video tool with your photo as the reference.
These are English versions of the scene directions our own maker runs. Every model reads a prompt a little differently, so expect to adjust a line or two; the fix list below covers the usual problems. Updated October 11, 2026.
Vertical 9:16 video, 8 seconds, cinematic, shallow depth of field. Use the uploaded photo as the character reference. Keep their face, skin tone, hair and identity exactly the same for the whole clip. They are the only person in sharp focus. Wardrobe: a dark shirt or knit with a jacket, fully dressed. Setting: a crowded house party at night in a wood-panelled room, large crystal chandeliers overhead, warm golden light, a light haze. Guests hold drinks and stay as soft, blurred shapes in the background; no guest shows a clear face. 0 to 3s: the person stands in the middle of the crowd with one hand above their eyebrows like shading their eyes from the sun, palm down, eyes visible, leaning forward and slowly searching the room from left to right. 3 to 6s: their eyes stop on the camera, they drop the hand, break into a delighted laugh and push through the guests toward the lens. 6 to 8s: they point straight at the camera while the guests behind them turn and raise their glasses, cheering. Hold the last frame. Camera: slow, steady move that stays on the person, waist-up framing. Audio: a low bass beat and crowd chatter. No singing, no lyrics. Avoid: face changing, a second sharp face, extra fingers, plastic skin, cartoon look, on-screen text, subtitles, logos, watermarks, the likeness of any real artist.
Why it is written this way
What each line of the prompt does
If you change the prompt, these are the lines to keep. Each one prevents a specific failure.
- Use the uploaded photo as the character reference…
- Pins who is on screen. Without it the model invents a person who only resembles you.
- They are the only person in sharp focus.
- The single most useful line. A second sharp face is what yours gets blended with.
- Wardrobe: a dark shirt or knit with a jacket…
- Says what to wear in positive words. Listing what not to wear can get a prompt refused by some models.
- Setting: a crowded house party… crystal chandeliers…
- The room is the trend. Wood panelling, chandeliers and a crowd with drinks are what people recognise.
- 0 to 3s… 3 to 6s… 6 to 8s…
- Timed beats stop the model from spending the whole clip on one action. Our first version put a walk first and the scan second; opening on the scan itself made a stronger clip.
- Audio: a low bass beat… No singing, no lyrics.
- Keeps generated vocals out. You add the real sound on the platform when you post.
When the first render is off
Fix a Burning Bridges AI video that came out wrong
| What you see | What to change |
|---|---|
| The face drifts or turns into someone else | Use a sharper, front-facing photo; keep the “only person in sharp focus” line; try a shorter clip. |
| Guests have clear, strange faces | Add “guests seen from behind or out of focus” to the setting line. |
| The hand never goes over the eyes | Put the scan in the first beat and say “like shading the eyes from the sun”: a flat hand at the brow can read as a salute. |
| Six fingers on the raised hand | Ask for “hand held flat, fingers together”. Re-roll: hands vary run to run. |
| The prompt is refused | Remove any real names, song titles and clothing you do not want; say what to wear instead. |
| Text or a logo appears on screen | Keep “on-screen text, subtitles, logos, watermarks” in the Avoid line, and do not ask for captions in the prompt. |
Or skip the prompt and just add a photo
The maker on our home page has all four moments built in. One photo, 8 seconds, vertical, with sound, from $9.99. A render that fails costs nothing.
Before you ask
Prompt questions
What is the best AI prompt for the Burning Bridges trend?
One that locks three things: who is on screen (your photo is the only clear face), where (a wood-panelled party room under crystal chandeliers, guests as soft background shapes) and what happens, split into timed beats (walk out of the crowd, hand over the eyes scanning the room, point at the camera). The builder on this page writes that prompt for four different moments.
Which AI do I paste the prompt into?
Any image-to-video tool that accepts a reference photo and a text prompt. The TikTok tutorials for this trend mostly name Kling; other image-to-video apps work the same way. Upload your photo as the reference or first frame, paste the prompt, choose vertical 9:16 and a length of 5 to 10 seconds.
Why does the face change halfway through my video?
Usually the photo, then the prompt. Use a sharp, front-facing photo with the face large in frame. Keep the line that says your photo is the only clear face and that guests stay blurred, because a second sharp face is what the model blends yours with. Shorter clips drift less: try 5 seconds before 10.
Does the prompt add the song?
No, and it should not ask for it. The prompt asks for a party beat and crowd noise with no vocals. Add the actual song from the TikTok or Instagram sound library when you post, so it is licensed there and your video is not muted.
Can I change the prompt to a footballer or a celebrity?
The prompt works on whatever photo you give it, and many clips in the trend do cast famous people. We do not recommend it: use your own photo or an adult who agreed. Platforms can remove AI clips of real people who did not consent, and it is not something this site makes.
Is the prompt free to use?
Yes. Copy it, change it, use it anywhere. What costs money is the rendering in whichever tool you paste it into. If you would rather skip the tuning, the maker on our home page has the scene built in and makes the clip from one photo, from $9.99.