How Do You Make a Cinematic AI Commercial? Shot-by-Shot Prompts for a Car, Headphones and Sports Ad

A cinematic AI commercial is three passes, not one prompt: a hero still of the product, motion generated from that still, and a polish pass for sound and edit. On VdoBloom a 2K hero still on Nano Banana Pro is 18 credits, and a 5-second 720p Seedance 2.5 clip is 126, so a twelve-shot spot is planned and budgeted before anything renders.

Updated October 2026.

The three-step workflow: hero still, motion, polish

A whole commercial in one prompt looks like a commercial and sells nothing. The reliable route keeps control at each stage.

Hero still. Generate the product in the exact composition, lens and light you want on image generation. Nano Banana Pro at 2K costs 18 credits and accepts reference images, so you can keep packaging and colour accurate.
Motion. Take that still into image-to-video and animate it with Seedance 2.5 for longer moves or Kling 3.0 when the shot must land on a planned last frame. The still is frame one, so the prompt describes only movement, camera and sound.
Polish. Cut the shots together, add music and sound effects, and finish with captions if the spot runs on a muted feed. The product demo tab shows the clean, functional version of the same job.

Every setting shows its credit cost before you generate, and failed renders are refunded. The prompts below are prompts to try, not outputs from a specific session.

Car commercial: six shots with prompts

Six shots cover a 20 to 30 second spot. Generate each hero still first, then animate.

Front three-quarter: “A silver coupe parked on wet asphalt at dusk, front three-quarter view, low camera at bumper height, 24mm wide lens, cool blue ambient light with a warm rim from the left, reflections on the paint.” — 5s, 720p, 16:9, slow dolly in.
Headlight macro: “Extreme close-up of a matrix headlight in a dark studio, individual LED elements visible, single hard key from above, deep shadow around the housing, 100mm macro lens.” — 4s, 720p, 16:9, locked off with a light sweep.
Wheel in motion: “A five-spoke alloy wheel rotating against blurred asphalt, camera tracking at hub height, late afternoon sun, motion blur in the spokes, 85mm lens.” — 5s, 720p, 16:9, side tracking shot.
Interior detail: “Driver hands settle on a stitched leather wheel, dashboard out of focus behind, soft window light from the right, 50mm lens, shallow depth of field.” — 5s, 720p, 16:9, slow push in.
Rear tracking: “A coupe drives away along a coastal road at golden hour, camera tracking behind at roofline height, long lens compression, dust lifting from the rear tyres.” — 8s, 1080p, 21:9, smooth follow.
Closing badge: “The rear badge catches a low sun against a dark body panel, tight framing, lens flare at the edge, 100mm macro, static shot.” — 4s, 1080p, 16:9, locked off.

Headphones: macro, rotation and light sweep

Keep the background plain and spend the detail on surface finish.

Earcup macro: “A matte black headphone earcup on a charcoal stone surface, brushed texture visible, soft top light, faint contact shadow, 100mm macro, three-quarter view.” — 4s, 720p, 1:1, slow lateral slide.
Slow rotation: “Headphones rotate once on a pale grey plinth against a seamless studio background, single large softbox from the upper left, even fill, 50mm lens.” — 8s, 720p, 16:9, orbit right.
Light sweep: “A narrow highlight travels across the headband and earcup as the light moves, background stays dark, controlled specular edge, 85mm lens.” — 5s, 1080p, 16:9, locked off.
Hands lifting: “Two hands lift the headphones from the plinth and hold them at chest height, soft window light from the right, neutral background, 50mm lens.” — 5s, 720p, 9:16, medium shot.
Worn in profile: “A person in a plain grey sweater wears the headphones in profile against a warm concrete wall, late afternoon side light, 85mm lens, shallow depth of field.” — 6s, 720p, 9:16, slow push in.

Sports gear: motion and grip shots that stay on-brand

Favour short bursts and clean backgrounds over chaotic action.

Shoe grip: “A running shoe plants on wet track surface, water displacing from the tread, low camera at ground level, hard side light, 100mm macro.” — 4s, 720p, 16:9, locked off.
Fabric stretch: “Close-up of a jersey fabric pulling across a shoulder, weave visible, soft diffused light from the left, 85mm lens, shallow depth of field.” — 4s, 720p, 1:1, slow push in.
Sprint start: “An athlete in plain training kit explodes from a starting position on a track, dust from the blocks, camera low and tracking alongside, overcast light.” — 6s, 720p, 16:9, side tracking.
Product in hand: “A hand grips a matte water bottle, condensation on the surface, gym background blurred, hard key from the right, 50mm lens.” — 4s, 720p, 9:16, medium close-up.
Impact detail: “A training glove hits a heavy bag, chalk dust in the air, camera tight on the knuckles, hard side light, dark background, 100mm macro.” — 4s, 720p, 16:9, locked off.

Our food and product ad prompt guide has more detail on surfaces, steam and pour shots if the spot needs them.

Camera, lens and lighting words to reuse

Concrete words change output. Vague ones get skipped.

Category	Words that work	What they do
Lens	24mm wide, 50mm standard, 85mm portrait, 100mm macro	Sets compression and how much background is visible
Move	slow dolly in, orbit right, tracking shot, locked off, tilt up	Gives the clip one clear camera behaviour
Light	hard key, large softbox, rim light, practical lamp, golden hour, overcast	Controls contrast and direction
Framing	extreme close-up, medium shot, wide shot, low angle	Sets how much of the product fills the frame

Use one move per clip and name the light direction. For a longer sequence, write the move as an order: “push in, then settle as the hand enters”. Our guide to camera movement prompts explains how each move reads on screen.

Keeping the product identical from shot to shot

Every shot in the spot should start from the same hero still or the same reference set. Re-generating the product from text for each shot is where labels change, proportions drift and colours shift.

One master still per product. Generate it once at 2K on Nano Banana Pro and reuse it as the first frame for every shot that shows the product clearly.
Use first and last frame. On Kling 3.0, supply the opening and closing stills when a move has to land on a specific composition.
Keep the prompt nouns fixed. Call the product the same thing every time and do not re-describe its colour or finish mid-prompt.
Match the light. Keep the key on the same side across shots so the cuts feel continuous.

The same identity problem is covered in AI video character consistency; product work is the easier version because a still can be reused exactly. If a shot needs a person holding the item, use a real photo with consent or a clearly synthetic presenter with disclosure.

Cutting the shots together and adding sound

Assemble in the timeline editor and keep the cut rhythm simple: an establishing shot, two details, one moving shot and a close. Let each shot run long enough to read; a 4-second macro needs a moment before the cut.

Sound carries more of the cinematic feel than people expect. A low room tone, a designed whoosh on the transition and one clear music bed do most of the work. Generate a voice-over in text to speech if the spot needs narration.

Check the finished spot at phone size before publishing: detail that reads on a monitor can vanish in a feed.

Frequently asked questions

How many shots does a 30-second AI commercial need?

Six to ten. The car spot in this guide uses six. More shots mean shorter cuts, which hides motion imperfections.

Which model should animate the stills?

Seedance 2.5 for moves longer than a few seconds, because it runs to 30 seconds, and Kling 3.0 when the shot needs first and last frame control.

How much does a hero still cost?

Nano Banana Pro is 18 credits at 1K or 2K and 24 at 4K. A 5-second 720p Seedance 2.5 clip is 126 credits, and the price for every setting is shown before you generate.

How do I stop the product changing between shots?

Reuse one master still as the first frame, keep the product nouns identical in every prompt, and match the light direction across shots. Reference-based models help when a person has to appear.

Can I use a real person in the ad?

Only with their consent, and disclose synthetic presenters where the platform or jurisdiction requires it. Check the rules that apply to your market before publishing.

Open image generation for the hero still, then image to video to animate it, picking Seedance 2.5 or Kling 3.0. New accounts start with 10 free credits and no card, one-time credits never expire, and there is no subscription needed. Pack options are on the pricing page.