Burning Bridges AI Video Prompt (Copy & Paste)

A good Burning Bridges AI video prompt does three jobs: it fixes the scene (chandelier-lit private dining room), the action (cup raise, crowd toast), and the camera (handheld push-in). Below are three prompts that do exactly that. Copy one, paste it into any image-to-video tool, attach your photo, and render.

Prompt 1 — The classic (safest starting point)

Handheld cinematic shot inside a dim, luxurious private dining room at night. The person from the photo sits at the head of a long table packed with well-dressed friends toasting and hyping them up. Crystal chandeliers glow warm overhead, candles and bottles on the table. The person raises a metal cup toward the camera and smiles, then sweeps their eyes across the room like they are looking for someone. The handheld camera slowly pushes in through the crowd toward the person, subtle natural shake. Warm chandelier light on faces, shallow depth of field, film grain, 35mm documentary style. Keep the person's identity, face and clothing exactly as in the photo. No text, no logos.

This is the version our generator uses by default. It keeps the person from your photo as the subject and rebuilds everything around them.

The line that matters most: the camera sentence. "Handheld camera slowly pushes in through the crowd, subtle natural shake" is what makes the clip read as part of the trend — remove it and you get a static portrait video that could be any template. If you only protect one sentence from editing, protect that one.

Prompt 2 — Close-up chandelier toast

Handheld close-up inside a dim private dining room at night. The subject's face and shoulders are lit by warm crystal chandelier light, friends holding up metal cups fill the frame slightly out of focus behind them. The subject laughs and raises a metal cup toward the camera, the crowd cheers, the handheld camera pushes in slowly. Shallow depth of field, warm golden tones, film grain, 35mm documentary style, 10 seconds, portrait 9:16.

Use this when the photo is a headshot and you want likeness front and center. It is also the most forgiving prompt if your photo is slightly soft.

The line that matters most: the light sentence. "Face and shoulders lit by warm crystal chandelier light" is doing the identity work here — when the face is the frame, the light on the face decides whether the render looks expensive or flat. Swap the light description and the whole clip changes character.

Prompt 3 — Pet edition

Handheld cinematic shot inside a dim, luxurious private dining room at night. A lone dog sits proudly at the head of a long table packed with well-dressed friends toasting and cheering. Crystal chandeliers glow warm overhead, candles and bottles on the table. The dog's ears perk up as the crowd raises metal cups, the handheld camera slowly pushes in. Filmic warm tones, shallow depth of field, 10 seconds.

Swap "dog" for cat, horse, or whoever your star is.

The line that matters most: the subject hold. "A lone dog sits proudly" pins the animal in place so the model spends its motion budget on the crowd and camera instead of inventing animal behavior. Animals wander in AI video unless you anchor them; that one verb, "sits," is the whole trick.

Make it your own

The three blocks share a skeleton: scene → subject action → camera → light → duration and ratio. To remix, replace exactly one layer per render:

1. Scene. Swap "private dining room" for a rooftop terrace, a club booth, a yacht cabin. Keep one location — "moves between rooms" produces chaos.
2. Action. Keep one gesture per ten seconds. "Raises the cup, then waves, then points" dilutes into a shrug. Pick the strongest beat and let the camera carry the rest.
3. Camera. The push-in is the trend signature, but "slow orbit around the person" and "pull-back revealing the crowd" both work. One move, continuous, no cuts.
4. Light and mood. "Warm chandelier glow" is the default; "cold blue hour through tall windows" gives you the somber cut some creators prefer.
5. Duration and ratio. Always state them. Tools default to shorter, wider clips when you leave these out, and your ten-second 9:16 upload quietly becomes a five-second 16:9.

Two practical rules keep remixes sharp: keep the whole prompt under about 120 words, and change one layer per render so you can tell what actually improved.

Five prompt mistakes that ruin the render

1. Writing a novel. Past ~120 words, models start trading your requirements against each other. The last thing you added is usually the first thing ignored.
2. Stacking actions. "He dances in, leans to the lens, raises the cup, waves, points at a friend" renders as a twitch. One beat per clip.
3. Omitting duration and aspect ratio. The single most common silent failure — the tool picks its defaults and your vertical trend clip comes out landscape and half as long.
4. Directing cuts. "Cut to," "then the camera switches to" — image-to-video models render one continuous take. Multi-shot language gets averaged into blur.
5. Describing the photo instead of the scene. "A man in the photo wearing a blue shirt" wastes half the prompt. The model already sees the photo; spend the words on what happens around him.

Where to paste it

Using our generator? The custom prompt box lives in Advanced Settings on the Generate page, pre-filled with Prompt 1. Using another tool? Paste it there, attach one clear photo, render — and if the first take misses, run the same prompt again before rewriting; these models have per-run variance. For the full photo rules and failure checklist, see the step-by-step tutorial.