how toburning bridgesdrake burning bridgeswhere she atwhere she at gesturedrake icemansilver shakerai music videoai pet videodrake trend

How to Make the Drake 'Burning Bridges' AI Video

'Where She At?', starring you or your pet — the 2026 trend, explained and built, start to finish.

Burning Bridge AI TeamUpdated October 11, 20267 min readMake yours →

Quick answer

One photo is all it takes. Upload it to the Burning Bridges generator and you get a 15-second vertical clip of you — or your pet — living the chandelier-party scene from Drake's 'Burning Bridges': silver shaker in your left hand, finger flicks at the lens, the collar grab, drinks poured into the crowd, and both 'Where she at?' hand-over-eyes covers, all on the beat with the song built in. You keep your own clothes and jewellery. Videos start at $9.90, and the clip is ready in about two minutes.

What you get

15 seconds, vertical 9:16, with the song and party sound built in
One photo in, one finished clip out — no prompts, no editing
Your own clothes and jewellery carry straight over
Pets run the full routine on two legs, own face and markings kept
Ready in about two minutes
Pay per video from $9.90 — no subscription

What the trend actually is

If your feed has filled up with one specific party — dark wood panelling, crystal chandeliers, a packed room in evening wear, and one person dead centre rapping with a silver cocktail shaker — you're looking at 'Burning Bridges'. The clip riffs on the track from Drake's 2026 album Iceman: the shaker up, the collar grab, drinks poured into the crowd, and the whole room moving on the beat behind you.

The moment everyone waits for is the 'Where she at?' cover: a hand raised flat over the eyes, scanning the room for someone who isn't there. It came out of a World Cup watch-party clip earlier this year — the kind of throwaway reaction that ends up carrying an entire meme.

The trend only broke out in late September 2026, and it's still climbing. It works because almost nothing is asked of you: no dancing skill, no camera time, no editing. One decent photo and the scene does the rest.

Pets welcome

The pet cuts might be the best part of this trend. Your dog or cat keeps its real face and markings, stands up on two legs, holds the shaker in a paw and raps along with its own mouth — same room, same choreography, same beat. If your pet has ever looked at you like it runs the house, this will feel like justice.

Pro tip: shoot the photo at your pet's eye level in even light, and include the full body. More of the animal in frame means more of the animal in the clip — bow ties and little chains carry over beautifully.

The fastest way

Open the Burning Bridges generator, drop in one clear front-facing photo and hit Create. The room, the crowd, the shaker and the beat are already directed — your photo just takes the lead role. A 15-second clip arrives in about two minutes, sized for TikTok, Reels and Shorts, song included.

It's pay per video: $9.90 for one, $19.90 for three, $29.90 for five — no subscription, and what you buy never expires. The exact cost of a video shows before anything generates, so there are no surprises at the end.

Try it — one photo is the whole job

Skip the setup entirely: the room, the crowd and the beats are already directed. You just bring a photo.

Make mine

Build it yourself

Want to understand how a clip like this is actually made — or cut your own version with your own tools? Here's the honest, technical version. It's four steps, and the last one is the only one that needs real patience.

Step 1 — Pull a depth pass

Map the room, lose the performer

Run the original clip's frames through a depth model like Depth Anything V2. You're stripping the original performer out and keeping the room itself: how the crowd compresses in the background, where the chandeliers hang, the way the camera pushes in. That map is what will drive your version — it carries the camera moves and the crowd's rhythm without carrying a single original frame.

Step 2 — Build the still

One frame that sets everything after it

Generate a single photorealistic frame of your subject matching the scene's light. The video model will spend the whole clip trying to look like this frame, so a bad still means a bad video. A prompt that works:

Still prompt

A photorealistic vertical 9:16 frame of the person from the reference photo, centre of a packed late-night party in a dark wood-panelled room lit by crystal chandeliers. They hype straight into the camera, a silver cocktail shaker raised in their left hand, the right hand pointing at the lens. Behind them, a dense crowd of guests lifts their drinks. Shot like an 85mm film still — shallow depth of field, creamy bokeh, warm halation, soft bloom off the chandeliers, fine grain. Keep their exact face, their own clothes and jewellery. No text.

Step 3 — Write the beat sheet

Choreography is a timestamped list

Watch the original at half speed and log every move with its time and lyric cue. A working version:

  • 0–2s — walk out of the crowd, shaker already up
  • 2–4s — finger flicks straight at the lens
  • 4–6s — arms-wide swagger, chin up
  • 6–8s — the collar grab
  • 8–11s — pour into the crowd's raised glasses
  • 11–13s — first 'Where she at?' cover, hand flat over the eyes
  • 13–15s — second cover lands on the hook, mouth still rapping

Skip this and the clip looks like a moving wallpaper. Write it and the model suddenly looks directed — because it is.

Step 4 — Drive and render

Depth map in, performance out

Feed the still, the beat sheet and the depth pass into a depth-conditioned video model — Seedance, Kling or whatever you can get access to — and render. Expect a few passes: hands, the shaker and the eye-covers are the usual trouble spots. When the render holds the beat sheet, you're done.

Common mistakes

Reusing frames from the original clip

Face-swapping onto the actual footage is the fastest route to a takedown — and it reads as fake instantly. Use the original only as a choreography reference and drive your own render.

Freestyling the choreography

Left to its own devices, a video model drifts and smooths everything into mush. The beat sheet is what turns motion into dancing.

Letting the shaker switch hands

The silver shaker stays in the left hand the whole way through. The moment it trades hands mid-clip, the illusion collapses.

Rushing the eye-cover

The 'Where she at?' moment is the whole point of the trend. Give each cover a full second on screen — don't let the model blink through it.

Starting from a weak photo

Sunglasses, busy group shots and harsh backlight are behind most failed generations. One person, face visible, even light. That's the entire brief.

Frequently asked questions

A short AI clip that drops you — or your pet — into the chandelier-lit party scene from Drake's 'Burning Bridges'. One photo goes in; a 15-second vertical clip comes out, with the moves and the sound already there.

A hand raised over the eyes, scanning the room. It's the moment from Drake's 2026 World Cup reaction that turned the whole party clip into a meme — and it's the beat everybody waits for.

One. A clear, front-facing shot with even light. Sunglasses and busy group photos are the two things that cause most of the failures.

Yes — dogs and cats run the full routine on two legs, keeping their own markings and face. A clear, full-body photo works best.

Yes. The switch swaps the person, not the wardrobe — your outfit, jewellery and look carry straight over into the clip.

Fifteen seconds, vertical 9:16, ready in about two minutes. Pay per video: $9.90 for one, $19.90 for three, $29.90 for five. No subscription, and what you buy never expires.

Yes. TikTok, Instagram and YouTube all ask you to tag realistic synthetic video — use the platform's built-in AI label when you post. It's one tap, and it keeps your account safe.

One tap

Make this on Burning Bridge AI

One photo, about two minutes, one 15-second clip with your name on the party.

Unofficial fan tool — not affiliated with, endorsed by or connected to Drake, OVO Sound, Republic Records or any record label. Song and artist names are used to describe the trend only.