How to make the Burning Bridges AI video
The Drake party scene, the silver cup, the hand over the eyes. Here is how people are putting themselves, their friends and their pets into it with AI, in three ways, with a prompt you can paste.
Get the promptSee the 3 methodsUpdated 2026-10-10. Independent guide, no tool is paying to be listed here.
What the trend is
"Burning Bridges" is a 2026 Drake track, reportedly from the album Iceman. The music video's second half is a packed party in a wood-panelled dining room under crystal chandeliers. Near the end Drake raises a flat hand over his eyes and scans the room, as if looking for someone. That move went viral on its own over the summer, and in early October 2026 creators started using AI to drop themselves into the scene, cup in hand, doing the same routine. The AI versions are what this site is about. For the backstory, see the Drake Burning Bridges trend explained.
Three ways to make it
- WAY 1 · EASIEST
One-photo template
Upload one photo, get a 20-second vertical clip with the routine already built in. No prompt. Usually a few dollars per video at sites like Starrd, Flowlio or LightX. Works with pets.
- WAY 2 · THE ORIGINAL LOOK
Swap yourself into the real scene
Higgsfield Genjutsu, Kling Motion Control or Seedance: a clip of the party scene as the motion reference, your photo as the identity, plus a prompt. 3 to 5 takes. Best for cartoon characters and the exact room.
- WAY 3 · NO AI
Film the gesture
Phone, friends, a cup. Act natural, then on the key line raise a flat hand over your eyes and slowly turn your head.
Way 2 step by step (Higgsfield, Kling or Seedance)
- Trim the reference. Cut 8 to 20 seconds from the party half of the official video. The wave at the camera and the hand-over-eyes move are the two beats every version keeps.
- Prepare your photo. One clear, front-facing shot, good light, waist up. Whatever you wear in the photo (chain, glasses, hat) carries into the clip, so dress for it.
- Choose character swap, not restyle. In Higgsfield pick Genjutsu and upload the video and your image. Restyle modes repaint the whole room; you only want the lead changed.
- Paste the prompt. Use the swap prompt below. Say which hand holds the cup, and say that only the central person changes. All prompt variants are on the prompt page.
- Generate 3 to 5 takes at the cheapest resolution, keep the cleanest, then re-run that one at higher resolution.
- Download and post. Most AI exports are silent; that is normal. Download details and format checklist.
The person in @Image1 replaces the man waving at the camera in @Video1, matching his position, pose, wave, cup and timing exactly. Keep everything else in @Video1 unchanged: the dining room, the chandeliers, the warm light, the other guests, the handheld camera movement and the cuts. Keep the face, hair and clothes from @Image1. Only the central performer changes; do not alter any background face.
The 20-second scene, beat by beat
If your tool takes a shot list, or you are writing your own prompt, this is the routine the templates reproduce.
- 0:00–0:02 Wide shot. You stand inside the crowd under the chandeliers, silver cup in your left hand, pointing out at the room.
- 0:02–0:05 Close shot. Right hand waves at the camera, then the cup comes up beside your head.
- 0:05–0:10 Medium shot, facing the lens. Cup at the waist, arms crossed, swaying with the beat while the mouth keeps rapping.
- 0:10–0:15 Three-quarter angle. One arm extends and the cup points at the dancing guests.
- 0:15–0:16 Quick finger flick toward the lens.
- 0:16–0:20 The signature move: right hand flat over the eyes, head turning slowly as if scanning the room, cup still up. Crowd cheers, glasses rise.
Which tool should you use?
| Tool | Good for | Watch out |
|---|---|---|
| Higgsfield (Genjutsu) | Where most viral versions came from, including the cartoon ones. Keeps the real room. | Free tier is limited and watermarked. Needs a trimmed reference clip. |
| Kling Motion Control | Clean motion transfer from a reference clip. | Follows only one performer, the largest in frame, so trim tight on Drake. |
| Seedance 2.5 | The model behind most one-photo templates. Good lip movement. | Direct use needs a still image plus a motion reference; more setup. |
| One-photo template sites | Fastest. Pets. No prompt. | Paid per video, usually 480p by default, fewer controls. |
| CapCut templates | Quick edits over existing footage. | Most do not generate a new scene; you are pasting a face, not performing. |
Why it keeps failing, and the fix
| Problem | Fix |
|---|---|
| The cup jumps from one hand to the other | Name the hand in the prompt, twice if needed: "silver shaker in the LEFT hand, never switches". |
| Drake's face or outfit leaks into your clip | Use a character-swap mode, not a restyle. Or run the clip through a depth-map model first so only motion is kept. |
| Background guests turn into copies of you | Add "only the central performer changes, background guests stay distinct". |
| Hand covers the mouth, lip sync breaks | Write "hand flat over the eyes only, mouth uncovered and still rapping". |
| Face does not look like you | Sharper photo, face larger in frame, no filters, no sunglasses. |
| Blurry output | Test at 480p, then re-run the good take at 720p or 1080p. |
Only use photos of yourself or people who have agreed. Do not make versions of kids or of people who would not want to be in it. Label the post as AI where the platform asks. The song belongs to its rights holders; keep it to the official in-app sound and do not sell the clip.
Burning Bridges AI video FAQ
What is the Burning Bridges AI video?
A short AI-made clip, usually about 20 seconds and vertical, that puts you, a friend, a pet or a cartoon character into the chandelier-lit party scene from Drake's "Burning Bridges" music video. The clip ends on the hand-over-eyes "searching the room" move that became a meme.
Which AI tool makes the Burning Bridges video?
Most early versions were made in Higgsfield's Genjutsu mode, which swaps a character into a reference video. Reference-to-video models such as Seedance 2.5 and Kling Motion Control also work. One-photo template sites use Seedance under the hood. CapCut templates mostly re-edit existing clips rather than generate new ones.
Do I need a prompt?
Only if you go the DIY route. One-photo templates need just a photo. For Higgsfield, Kling or Seedance, copy a prompt from the prompt page and adjust the cup hand and the ending.
How many tries does it take?
Expect 3 to 5 generations before one looks clean. Common misses: the cup jumps hands, a background guest turns into a copy of you, or the hand covers the mouth instead of the eyes.
Can I do it with my dog or cat?
Yes. Use a photo of the pet alone, standing or sitting with the face clearly visible, and use the pet prompt. Results are more hit or miss than with people.
Is it free?
Higgsfield, Kling and most video models have a free tier with daily credits and a watermark. One-photo templates usually cost a few dollars per video. Nothing on this site costs money.
Is it legal to post?
The song and the original video belong to Drake's label, and we are not lawyers. Keep it to a personal parody clip, do not sell it or use it in ads, only use photos of yourself or people who agreed, and label the post as AI-generated where the platform asks.