Veo 3.1 prompts in the order Google recommends
Google's Ultimate prompting guide for Veo 3.1 puts the parts of a shot in one order, and every card here follows it:
- Cinematography: framing and camera move, such as "slow tracking shot".
- Subject: who or what, in detail you could draw.
- Action: one thing that happens inside 4 to 8 seconds.
- Context: the place and the time of day.
- Style and ambiance: light, palette, lens and mood.
Then write the sound. Put speech in quotes and name the speaker (he says, "…"), and give effects and background their own lines that start "SFX:" and "Ambient noise:". Google's Veo documentation for the Gemini API splits audio into the same three kinds.
To leave something out, describe the scene without it: "a quiet beach with no one on it", not "no people". The studio has one prompt box and no negative field.
To assemble one from parts, build your own Veo prompt slot by slot or write the same shot as a JSON prompt.
The limits that change a Veo 3.1 prompt
Credits are 160 a second at 720p or 1080p and 240 at 4k. Choosing 1080p or 4k locks any clip to 8 seconds, so 1,280 or 1,920 credits.
| Mode (studio name) | You attach | Ratio | Length | Credits, 720p or 1080p |
|---|---|---|---|---|
| Text to video (Describe) | Nothing | 16:9 or 9:16 | 4, 6 or 8 s | 640 to 1,280 |
| Image to video (Animate) | 1 image | 16:9 or 9:16 | 4, 6 or 8 s | 640 to 1,280 |
| Reference images (Reference) | Up to 3 images of one subject | 16:9 only | 8 s | 1,280 |
| First and last frame (Transition) | 2 images | 16:9 or 9:16 | 8 s | 1,280 |
| Extension (Extend) | 1 Veo clip, 720p | Same as the clip | Adds about 7 s | 1,280 per call |
What no prompt can change on Veo 3.1:
- No square, 4:5 or 21:9 frame, and no silent clip.
- No seed, so a take cannot be repeated exactly.
- No video edit; that runs on Wan 2.7 or HappyHorse 1.1.
- Every clip carries Google's SynthID watermark. For requests from the EU, UK, Switzerland and MENA, Google allows only adults in clips.
Text to video: scenes with written sound
Single shots with no dialogue. Sound gets its own lines, because Veo adds audio to every clip whether you ask or not.
-
Noodle stall at the rush
16:9 · 8 s · 1,280 crSlow tracking shot along a night-market noodle stall, a cook tossing noodles in a flaming wok, steam rising into the string lights. Warm tungsten glow, 35mm film look. SFX: the wok hisses, a ladle clangs. Ambient noise: market chatter. -
Lamp room at midnight
9:16 · 8 s · 1,280 crSlow tilt up from a close-up: an elderly lighthouse keeper polishes the glass lens by lantern light as the beam sweeps past. Cold blue night, a solitary mood. SFX: cloth squeaking on glass. Ambient noise: wind on the tower. -
Ferry crossing at dawn
16:9 · 6 s · 960 crWide shot from the stern of a ferry crossing a grey fjord at dawn, gulls over the wake, one passenger in a yellow slicker at the rail. Muted palette, low mist, documentary realism. SFX: the engine's steady throb. Ambient noise: wind and water. -
Croissants in, 5 a.m.
9:16 · 8 s · 1,280 crHandheld medium shot in a bakery at 5 a.m.: a baker brushes egg wash over a tray of croissants and slides it into a deck oven, flour drifting through the lamp light. Golden light, soft grain. SFX: the oven door clanking shut. Ambient noise: a radio playing low. -
Rooftop in the rain
16:9 · 8 s 4k · 1,920 crCrane shot rising over a city rooftop in heavy rain at dusk, a woman under a yellow umbrella watching the skyline light up. Neon reflections, blue and magenta palette. SFX: rain drumming on the umbrella. Ambient noise: distant traffic hiss. -
Mountain hut, first snow
16:9 · 8 s · 1,280 crSlow push-in toward a stone mountain hut at blue hour as the first snow falls, one window glowing amber. Wide-angle lens, calm and remote. SFX: a wooden door creaks shut. Ambient noise: soft wind, no music. -
Crate digging
9:16 · 4 s · 640 crPOV shot flipping through a crate of records in a vinyl shop, fingers stopping on a faded orange sleeve, dusty afternoon sun. Warm, nostalgic, shallow depth of field. SFX: sleeves knocking together. Ambient noise: jazz playing softly. -
Espresso machine, first drops
16:9 · 4 s · 640 crMacro shot of an espresso machine's chrome group head as the first dark drops hit a small white cup, crema blooming in a warm backlight. Rich brown and chrome palette. SFX: the pump buzzing, then a steady drip.
Dialogue: a character says a line
Quote the exact words and name who says them, as Google's guide does. Keep each line short enough to say naturally inside the clip.
-
The stall owner calls an order
9:16 · 6 s · 960 crMedium close-up of a noodle-stall owner in a white cap, grinning through the steam. He says, "Two spicy, one mild, coming right up!" Night-market bokeh, warm string lights. SFX: a wok sizzling off screen. -
The keeper's log
16:9 · 8 s · 1,280 crClose-up of a lighthouse keeper writing in a leather logbook, the beam sweeping the window behind him. He murmurs, "Wind from the north. No ships tonight." Lantern light, deep shadows. SFX: the nib scratching. -
Two strangers at the rail
16:9 · 8 s · 1,280 crTwo-shot on a ferry deck at dawn: a young man offers his thermos cup to an older woman at the rail. He says, "Coffee? It's still hot." She smiles and says, "Only if you tell me where you're headed." Cold mist, soft light. -
The record-shop pick
9:16 · 6 s · 960 crMedium shot of a record-shop clerk holding an LP up to the camera. She says, "This one changed my life. Don't tell anyone." Warm afternoon light, gig posters behind her. Ambient noise: a record crackling on the shop turntable. -
Coach at the net
16:9 · 8 s · 1,280 crLow-angle medium shot of a tennis coach at the net on a clay court, pointing her racket at the lens. She calls out, "Watch the ball, not the player. Again!" Late sun, orange dust in the air. SFX: a ball bouncing on clay. -
Radio check from the hut
16:9 · 8 s · 1,280 crClose-up of a hut warden by a wood stove, keying an old handheld radio. She says calmly, "Base, this is the hut. Two hikers in, safe and dry." Firelight on her face, snow at the window. SFX: a burst of static.
Timestamp prompts: four beats in eight seconds
Google's Veo 3.1 guide splits an 8-second clip into four timed beats, each with its own shot and sound. Set the clip to 8 seconds first. At Veo's 24 frames per second, each beat is 48 frames.
Count frames per beat with the storyboard frame counter
-
One bowl, start to finish
16:9 · 8 s · 1,280 cr[00:00-00:02] Wide shot of a night market, one noodle stall glowing. SFX: chatter. [00:02-00:04] Close-up: noodles hit a flaming wok. SFX: a loud sizzle. [00:04-00:06] The cook flips them into a paper bowl. SFX: a ladle tap. [00:06-00:08] A customer takes the bowl, steam rising. SFX: coins on the counter. -
Match point on clay
16:9 · 8 s · 1,280 cr[00:00-00:02] High wide shot of a clay court at golden hour, a rally under way. SFX: ball strikes. [00:02-00:04] Low tracking shot as a player slides into a backhand. SFX: shoes skidding. [00:04-00:06] Slow-motion close-up: the ball clips the net cord. SFX: a soft tick. [00:06-00:08] It drops over; she raises a fist. Emotion: quiet triumph. -
Arriving at the hut
9:16 · 8 s · 1,280 cr[00:00-00:02] Wide shot of two hikers climbing to a stone hut at dusk. SFX: boots crunching. [00:02-00:04] Close-up: a gloved hand knocks on the door. SFX: three knocks. [00:04-00:06] The door opens on firelight and the warden waves them in. [00:06-00:08] They pull off their hats by the stove. Ambient noise: a crackling fire. -
The ferry comes in
16:9 · 8 s · 1,280 cr[00:00-00:02] Aerial view of a ferry cutting through fog toward a harbour. SFX: a deep horn. [00:02-00:04] On deck, passengers pull on coats. Ambient noise: wind. [00:04-00:06] The ramp lowers onto the quay. SFX: chains and a clank. [00:06-00:08] A cyclist rolls off into pale sunlight. SFX: tyres on wet planks.
Animate a still (image to video)
Attach one image in Animate mode and write only what changes: motion, camera and sound. Match your image's ratio: Veo letterboxes any other shape, and the black bars stay in the clip.
-
Product still: espresso machine
16:9 · 6 s · 960 crThe espresso machine in the image comes to life: steam bursts from the wand, coffee streams into the cup and a glint of light slides along the chrome. Slow push-in. SFX: the wand's hiss, the pump humming. -
Portrait: the keeper looks up
9:16 · 6 s · 960 crThe man in the portrait blinks, turns toward the window and a faint smile forms, wind stirring his collar as the lighthouse beam passes over his face. Locked-off camera. Ambient noise: waves and a distant gull. -
Landscape: snow over the hut
16:9 · 8 s · 1,280 crSnow begins to fall across the photo, smoke drifts from the hut's chimney and low clouds slide over the peaks. Slow push-in toward the lit window. SFX: timber creaking. Ambient noise: wind over the ridge. -
Illustration: rain on the rooftop
9:16 · 8 s · 1,280 crBring the illustration to life, keeping its painted texture: rain streaks down, neon signs flicker on and the figure's umbrella tips in a gust. A slow drift to the right. SFX: rain on gravel. Ambient noise: a distant city hum. -
Still life: courtside bench
16:9 · 8 s 4k · 1,920 crA breeze picks up across the photo: a tennis ball rolls off the courtside bench, leaf shadows flicker over the clay and the net sways. Slow push toward the racket strings. SFX: the ball dropping onto clay. Ambient noise: wind in the trees.
Reference images: keep one subject on screen
Reference mode takes up to three images of one person, character or product. Veo has no image tag, so describe the subject in words that match the pictures. On Playcut it is 16:9 and 8 seconds only.
-
A person from three photos
16:9 · 8 s · 1,280 crThe woman with the silver bob and green wool coat walks along a ferry deck at dawn, stops at the rail and turns to the camera. Medium tracking shot, overcast light, muted colour. SFX: footsteps on steel. Ambient noise: engine and wind. -
A character in a new place
16:9 · 8 s · 1,280 crThe round blue robot with one lamp for an eye rolls between the crates of a vinyl shop and nudges a record sleeve out with its arm. Low tracking shot, warm lamp light, playful. SFX: a small motor whirring. -
The espresso machine at work
16:9 · 8 s · 1,280 crThe brushed-steel espresso machine with walnut handles sits on a marble café counter at sunrise as hands lock in the portafilter and a shot pours. Slow orbit, warm window light. SFX: the handle clicking in, the pump humming. -
The fountain pen writes
16:9 · 8 s · 1,280 crThe black fountain pen with the gold clip signs a short name on cream paper, then is capped beside a glass ink bottle. Macro tracking shot on the nib, desk-lamp light. SFX: the nib scratching, the cap clicking shut.
First and last frame: describe the move between
Transition mode takes a start and an end frame and always runs 8 seconds. The prompt describes what happens between them: what moves, how the camera follows, what you hear.
Nano Banana prompts for making the two frames
-
Dusk to rush hour at the stall
16:9 · 8 s · 1,280 crThe camera holds still as dusk turns to night: the stall's shutter rolls up, string lights blink on and the empty stools fill. SFX: a shutter rattling, a gas burner igniting. Ambient noise: the market getting louder. -
Dough to loaf
9:16 · 8 s · 1,280 crOne slow push-in as the ball of dough on the bench rises, splits along its score and bakes into the golden loaf on the rack, flour dust swirling. SFX: a low whoosh, then an oven door thudding shut. -
Dry roof to downpour
16:9 · 8 s · 1,280 crTime-lapse feel from first frame to last: clouds roll over the rooftop, the sky darkens and rain sweeps the gravel as the neon signs switch on. The camera rises slowly. SFX: thunder building far away. -
Blank page to finished letter
16:9 · 8 s · 1,280 crOverhead locked-off shot: the fountain pen glides across the blank page, lines of ink spreading until the letter matches the final frame, the window light shifting from morning to afternoon. SFX: the nib scratching steadily.
Extend a Veo clip you already like
Extend adds about 7 seconds to a 720p clip Veo made in the last two days, up to 20 times, on a source of 141 seconds or less. Describe only what happens next. Google notes a voice carries on only if it is audible in the source's last second.
-
The next order
16:9 · +7 s · 1,280 crContinue the shot: the cook tips the noodles into a bowl, slides it across the counter and turns to the next customer. The camera drifts left along the stall. Ambient noise: the same market chatter. -
Down the spiral stairs
9:16 · +7 s · 1,280 crContinue: the keeper sets down his cloth, lifts the lantern and starts down the spiral stairs, the camera following from above. SFX: boots on iron steps. -
The harbour appears
16:9 · +7 s · 1,280 crContinue the crossing: the fog thins, a harbour of red boathouses comes into view and the passenger at the rail lifts a hand to wave. Same wide stern angle, same muted palette. SFX: the horn sounds twice. -
The conversation goes on
16:9 · +7 s · 1,280 crContinue the conversation: the older woman takes the cup, sips and says, "I'm going home. First time in twenty years." The young man nods slowly. Same two-shot, same dawn light. -
One more point
16:9 · +7 s · 1,280 crContinue the rally: the opponent lunges for a drop shot and scoops it back, and both players freeze as the ball hangs in the air. The camera stays low behind the baseline. SFX: shoes skidding, a sharp volley.
Pasting a Veo 3.1 card into the video studio
Veo 3.1 is already selected when you open the Playcut video studio. Then:
- Pick the mode: Describe, Animate, Reference, Transition or Extend.
- Match the tag: set the ratio, resolution and length the card names.
- Attach what the mode needs: one image, up to three references, two frames, or the Veo clip to extend.
Credits are held at submit and refunded if the job fails. Hobby's 500 monthly credits do not cover one 4-second Veo clip (640 credits). To rough out a shot for less, draft it on Seedance 2.0 from 10 credits a second or Grok Imagine at 20.
Every setting is on the Veo 3.1 modes, limits and credit cost page. For jobs tagged across several models, see the AI video prompts sorted by job.
If a card goes wrong, fix a Veo prompt that misfires symptom by symptom. For the rest of the studio's video tools, see the Playcut AI video generator.