Veo 3.1 Prompt Guide: 25 Examples That Actually Work
A tested Veo 3.1 prompt formula with 25 verbatim examples across cinematic ads, brand films, product shots, and experimental styles. Result observations included.
The Veo 3.1 prompt formula is: Subject + Action + Environment + Camera + Lighting + Mood. That six-element structure is the difference between a clip that looks filmed and one that looks generated. Veo 3.1 is the most cinematography-aware model available right now, it responds to lens language, lighting setups, and dialogue cues in ways that earlier models and most competitors don't. Get the formula right and you'll get output you can drop into an edit. Skip one element and you'll spend credits chasing results.
TL;DR
- The formula: Subject + Action + Environment + Camera + Lighting + Mood. All six elements, every prompt.
- What makes Veo 3.1 different: native dialogue support, audio sync, cinematography vocabulary, motion blur control. You can write like a DP.
- 25 tested prompts below, grouped by category, with result observations and variant notes. These came out of 8frame's Veo 3.1 generation queue in May 2026.
Why Veo 3.1 Prompts Work Differently
Most video models treat your prompt as a loose description and fill in the gaps. Veo 3.1 treats it as a shot list. The difference shows up in three places.
Cinematography vocabulary. Tell Veo 3.1 "35mm lens, rack focus from foreground to subject, slight handheld shake" and it executes that. Not as an aesthetic accident but as an actual camera move. Earlier models and most competitors soften these instructions into vague stylistic shifts. Veo 3.1 takes them literally enough that it's worth writing prompts the way a cinematographer would talk.
Lighting language. "Golden hour backlight, two stops underexposed, rim light from screen right" produces a specific result, not just a "warm" look. This is the main reason Veo 3.1 wins on cinematic brand work in our best AI video generator comparison for 2026.
Dialogue and native audio. Veo 3.1 is one of the few models that can sync lip movement to scripted dialogue. You include the spoken line in the prompt, in quotes, and the model generates the audio and mouth movement together. It's not perfect on every take, but it's functional enough that you're not always dubbing in post.
Motion blur control. You can specify shutter speed equivalent to control how motion blur reads. "1/50 shutter equivalent, 24fps" gives you that film-motion cadence. "1/500 shutter" freezes action for a product-shot feel.
The 6-Element Veo Prompt Formula
Each element earns its place in the prompt. Dropping one leaves Veo filling in that dimension with a default, and the defaults are rarely what you'd have chosen.
1. Subject
Be specific. Not "a woman" but "a woman in her 40s, dark hair, wearing a linen blazer." Veo doesn't have reference conditioning the way Seedance does, so your text description is doing all the work on appearance. The more specific you are, the more consistent the output across takes.
Weak: A man walks into a cafe. Strong: A man in his 30s, close-cropped beard, charcoal sweater, walks into a sunlit Paris cafe.
2. Action
State the action and the pacing. "Walks slowly" and "strides across" produce different energy. Include the direction of movement when it matters for camera composition: "moves toward camera" vs. "walks past camera left to right."
3. Environment
Give the setting enough detail to constrain the visual. Architecture, time period, weather, surface textures. If you don't specify, Veo picks generic. "An industrial loft with exposed steel beams, concrete floor, afternoon light coming from north-facing windows" beats "a modern interior."
4. Camera
This is where Veo 3.1 earns its cost premium. Specify: lens focal length (24mm wide, 85mm portrait, 135mm telephoto), movement type (static, slow push-in, tracking shot, handheld, crane up, aerial), and depth of field (shallow, deep, rack focus). You can also name shot size: wide shot, medium, close-up, extreme close-up.
5. Lighting
Name your light source, direction, and quality. Hard light vs. soft light. Natural vs. practical. "Overcast diffused daylight, no shadows" reads differently from "late afternoon window light, long shadows." You can reference film lighting shorthand: "rembrandt lighting from camera left" works.
6. Mood
One or two adjectives that aren't just visual. "Tense, unsettled" vs. "warm, nostalgic." This influences pacing, color temperature, and the model's micro-decisions on how subjects hold themselves and how the camera settles on cuts.
25 Tested Veo 3.1 Prompts
These were generated on the 8frame canvas in May 2026 using Veo 3.1 at default 4K settings, 5-second clip duration, 24fps unless noted. Result observations are one to two sentences describing what actually came out.
Cinematic Ads (5 Prompts)
Prompt 1
A woman in her 30s with wet hair steps out of a sleek black SUV at night, black dress, city lights reflected on the wet asphalt. Camera starts at low angle looking up, slow crane rise to eye level as she exits. Hard rim light from car interior, city bokeh behind her. Confident, cinematic.
Result: Clean crane move, practical interior light read correctly, bokeh was sharp without smearing. Hair texture held up across the clip. Strong for luxury automotive or fashion brief.
Variant: Swap crane to handheld push-in for a more urgent, editorial feel.
Prompt 2
Close-up of a hand placing a glass perfume bottle on a white marble surface, slow motion. Water droplets catch studio key light from above, soft fill from camera right. Static macro shot, extreme shallow depth of field, focus on the cap. Clean, minimal, aspirational.
Result: Water droplets rendered with real refraction. Depth of field was accurate to the focal length. Slow motion was smooth, no frame-rate artifact. Ready for use as a social cut for a fragrance brand.
Variant: Add "bottle sweating with condensation" for a more premium look.
Prompt 3
A coffee cup sits on a windowsill in a Paris apartment, steam rising. Morning light through sheer curtains, overcast sky outside. Slow rack focus from the window droplets to the cup. Medium close-up. Quiet, domestic, unhurried.
Result: Steam simulation was one of the best we've seen from Veo. The rack focus landed on frame. Morning light felt genuinely grey-Parisian rather than generic blue. The curtain material moved with subtle realism.
Variant: Add "a hand wraps around the cup from screen left" to bring human presence in.
Prompt 4
An athlete in red running shoes sprints through a forest trail at dawn, camera tracking from behind at ground level, slight handheld shake. Dappled early light through pine trees, rim light catching the shoes. Fast, kinetic, determined. 60fps.
Result: The 60fps specification held, motion blur was correct for speed. Ground-level framing gave the clip real energy. The shoe got highlighted on almost every stride, which read well for a product-forward edit.
Variant: Cut to 24fps for a moodier, less commercial feel.
Prompt 5
Two friends laugh at a small restaurant table at night, warm candlelight between them, bokeh of other diners behind. Medium two-shot, static camera. One friend says "This is exactly what I needed." Lip sync on speaker. Warm, intimate, real.
Result: Dialogue prompt worked on the second take. First take had slight mouth-movement lag, second was clean enough to use without dubbing. Candlelight handling was the best of any social-setting prompt we've tried in Veo.
Variant: Use a wider shot to show the full restaurant environment for a lifestyle brand context.
Brand Films (5 Prompts)
Prompt 6
A craftsman's hands shape wet clay on a pottery wheel in a clay-dusty studio, slow close-up. Diffused window light from the left, no fill, shadows on the right side of the clay. Medium close-up on hands and clay, slight slow push-in over 5 seconds. Quiet, skilled, meditative.
Result: Hand texture was exceptional. The clay surface read as genuinely wet. The light falloff on the unlit side gave the frame real depth. This one required zero post.
Variant: Pull back to a medium shot to reveal the craftsman's face in the background, out of focus.
Prompt 7
Aerial shot slowly rising above a field of yellow canola flowers at golden hour, slight wind movement in the crop. Wide lens, camera drifting upward from 10 feet to 30 feet over 5 seconds. Warm backlight, long shadows across the field. Expansive, alive, elemental.
Result: The aerial drift was smooth and motion felt intentional rather than floaty. Golden-hour light was the correct warm amber, not the oversaturated orange most models default to. Wind-driven crop movement was subtle but present.
Variant: Add a farmhouse in the background to anchor the shot for an agricultural or food brand.
Prompt 8
A scientist in her 50s with grey-streaked hair looks through a microscope, then looks up directly at camera with a small smile. Lab setting, blue-white fluorescent ambient, warm practical desk lamp on her face. Medium close-up. Confident, pioneering, grounded.
Result: The direct-to-camera look was clean. The transition from downward gaze to lens contact was natural, not robotic. The dual-light setup (fluorescent ambient vs. warm practical) read as intended. Good for pharma or deep-tech brand work.
Variant: Extend to two characters by adding a colleague in the background, blurred.
Prompt 9
Time-lapse of a city street at blue hour, lights turning on in building windows as pedestrians blur past. Camera static, mounted high on a tripod. Clean compression from 85mm equivalent, wide frame. Cinematic grain. Transitional, contemplative.
Result: The blue-hour gradient was accurate and held throughout the clip. Building window illumination came on in a realistic stagger rather than all at once. Pedestrian blur looked genuinely shot on a long exposure.
Variant: Specify a single lit window to cut to, for a storytelling transition in a longer edit.
Prompt 10
A woman in a white linen shirt stands at the edge of a dock at sunrise, her back to camera, looking across still water. Camera slowly pushes in from behind, from wide shot to medium. Mist on the water, soft pink-gold sky. Still, introspective, hopeful.
Result: Mist on water is a Veo strong suit. This was the best mist simulation we got in the test batch. The push-in had appropriate speed, not rushed. Good emotional anchor shot for a wellness or financial brand.
Variant: Add a boat appearing in the distance mid-clip for a sense of journey beginning.
Character Moments (5 Prompts)
Prompt 11
A teenage boy sits alone on a school bleacher at dusk, backpack at his feet, staring at his phone. Overcast grey sky, desaturated colors, parking lot lights just starting to flicker on. Camera at eye level, static. He says quietly, "I thought they'd show up." Slight lens flare from the nearest parking light. Lonely, real, unresolved.
Result: Dialogue line landed on second generation. The flicker of the parking lot light added exactly the kind of micro-detail that makes a scene feel observed rather than staged. Desaturation read correctly without going monochrome.
Variant: Remove dialogue and hold on the phone screen (dark, no reply) for a silent cut.
Prompt 12
An elderly man plays chess alone at a corner cafe table, late afternoon, no other customers. Warm window light from behind, cross-light from a pendant lamp above. He picks up the black queen, holds it, sets it down. Close-up on hands and board, then pull back to reveal empty chairs across from him. Melancholic, precise, still.
Result: Object handling (picking up and placing the chess piece) was the most convincing we've seen from Veo on a small-scale physical interaction. The pull-back to reveal empty chairs had genuine weight. Good for a story-driven brand spot.
Variant: Cut to the man looking toward the door to layer in anticipation.
Prompt 13
A 7-year-old girl runs down a long hallway toward a grandmother who crouches with open arms at the end. Tracking shot from behind the girl at low angle, camera rising as she reaches the end. Warm interior light, slightly warm grade. Joyful, uncomplicated, kinetic.
Result: The tracking move was one of the smoothest we got. The child's run was physically accurate (not the stiff stride you see from lesser models). The grandmother character was consistent to the end of the clip. Strong emotional payoff at the hug.
Variant: Shoot at 60fps and slow to 40% in post for a cinematic memory-like quality.
Prompt 14
A musician in a dark practice room plays a single chord on an upright piano, pauses, listens. Practical lamp on the piano, deep shadows in the room. Camera at piano level, slightly side angle. He writes something in a notebook. Focused, searching, solitary.
Result: The piano keys showed actual depression physics when the chord was played. The notebook writing gesture was clean. Low-key lighting held well, no muddy blacks. Good for a music or creative brand.
Variant: Add ambient room sound cue to the prompt for Veo to generate audio texture.
Prompt 15
Two colleagues walk side by side down a glass-walled office corridor, mid-morning light, one carries a coffee. They're mid-conversation but we can't hear them. Tracking shot alongside them, camera at shoulder height. Clean corporate minimalism, glass, steel, light wood. One of them laughs. Professional, forward-moving, candid.
Result: The silently-talking gesture read as genuinely mid-conversation, not posed. The tracking shot stayed steady without mechanical precision. The laugh was natural. Clean for a B2B or enterprise brand film.
Variant: Add specific dialogue on the laughing character for a slightly warmer, more personable cut.
Product Hero Shots (5 Prompts)
Prompt 16
A matte black wristwatch rotates slowly on a black velvet surface, studio light from above-left at 45 degrees, soft fill from below. Macro lens, extreme shallow depth of field, face in sharp focus, lugs just falling out. 360-degree rotation over 5 seconds. Precise, luxurious.
Result: The reflective watch face didn't smear or blow out, which is where most models fail on black metallic surfaces. The rotation was consistent at the stated speed. The velvet texture was visible without pixelating. Usable as a product hero shot with no retouching.
Variant: Add "a single water drop falls onto the watch face and beads off" for a material-story cut.
Prompt 17
A minimal white sneaker sits on raw concrete, camera at shoe level, slow rise to low aerial looking down. Natural overcast light, no shadows, clean background. Slight dust particle in air. The shoe is the only object in frame. 5 seconds. Fashion-forward, product-focused.
Result: The low-to-aerial move was smooth. The concrete texture was gritty and real without overwhelming the product. Overcast light was flat as requested, which is what you want for a product cut that's going to be color-graded. No artifacts on the shoe silhouette.
Variant: Add a person's foot stepping into the shoe at the start for lifestyle context.
Prompt 18
A glass bottle of olive oil is tipped slowly over a bowl of torn bread, oil pouring in a smooth arc. Natural kitchen window light from the left. Close-up on the pour, oil catching the light. Slow motion, 1/500 shutter equivalent to freeze the stream edge. Warm, tactile, Mediterranean.
Result: Liquid simulation was excellent. The oil stream refracted correctly. Bread texture was convincing. The 1/500 shutter specification slowed the edge of the pour exactly as intended. Strong for a food brand or DTC olive oil.
Variant: Add steam from warm bread to suggest it was just baked.
Prompt 19
A skincare serum bottle sits on a wet stone surface in a bathroom at golden hour. Camera at counter level, slow push-in to extreme close-up on the dropper tip. Water droplets on the stone, warm backlight rim-lighting the bottle. Clean, minimal, clinical-but-warm.
Result: The wet stone surface texture held throughout the push. Rim-lighting on the bottle glass was handled well with no smearing. The dropper tip at extreme close-up showed glass-lens bokeh on the background that looked authentic. Usable for a premium skincare brand.
Variant: Show a hand picking up the bottle at the end of the push for a use-context cut.
Prompt 20
A pair of over-ear headphones hangs on a simple wooden peg against a white plaster wall, soft morning light from a nearby window. Camera static, medium close-up. Wind from outside moves a thin curtain slightly in the left edge of frame. Considered, quiet, design-forward.
Result: The curtain micro-movement in frame edge added exactly the kind of life that separates a clinical product shot from a brand film. The headphone materials (leather pad, metal frame) were distinct and readable at medium close-up. Good for a consumer electronics or audio brand.
Variant: Add a morning ambient sound cue so the audio and visual mood match.
Stylized / Experimental (5 Prompts)
Prompt 21
A 1970s film grain look: a lone diner waitress refills a coffee cup under fluorescent light, late night, empty stools. Handheld, slightly shaky. Pushed-grain 16mm aesthetic, faded colors, green fluorescent cast. Medium shot, static-then-slight-drift. Lonely, Edward Hopper, American.
Result: The 16mm aesthetic was one of the cleanest film emulations we've run through Veo 3.1. The green fluorescent cast was accurate to era without tipping into parody. The slight camera drift felt genuinely handheld. Strong for a film-adjacent or nostalgia brand.
Variant: Add a jukebox in background bokeh and a specific song title as an audio cue.
Prompt 22
Hyper-stylized fashion: a model in an oversized coat walks slowly through a white fog-filled studio, high-key lighting, no shadows. Camera moves in slow orbit around the subject. Colors muted to near-monochrome except the coat which is saturated red. Minimal, graphic, editorial.
Result: The selective color hold (muted everything except red coat) worked on the first generation. The orbit move was stable. The fog fill diffused correctly without obscuring the subject. Ready for a high-fashion editorial campaign.
Variant: Add a second coat color in the second half of the clip for a palette-shift cut.
Prompt 23
Top-down flat-lay of a morning desk setup: journal, pen, ceramic mug, dried flowers. Camera descends slowly from 4 feet to 18 inches, perfectly vertical. Natural side-light from the left, soft shadows. Objects are still. Warm, analog, intentional.
Result: The perfectly-vertical descent is a technical request Veo 3.1 handles better than competitors. No lateral drift. The dried flower textures were detailed enough that individual petals were readable. Clean and usable for a stationery, wellness, or productivity brand.
Variant: Have a hand enter frame and open the journal mid-descent for a narrative beat.
Prompt 24
Abstract: colored ink drops fall in slow motion into a water tank, blooming in soft tendrils. Three colors: cobalt blue, warm amber, muted green. Camera front-on, medium shot, lit from behind to backlight the tank. No human subjects. 60fps. Meditative, scientific, beautiful.
Result: The ink diffusion physics were convincing and the backlit-tank setup read as requested. The three color interactions blended without muddying. The 60fps held without frame artifact. Good for a transition card, title sequence, or brand abstraction.
Variant: Add "one drop at a time, 2 seconds apart" to control pacing.
Prompt 25
Night cityscape: a neon-lit street in Tokyo seen from above, rain-slicked pavement reflecting signs. Camera slow drift from left to right at rooftop level. Long exposure motion blur on pedestrians, cars as light trails. Deep blues, pinks, greens from neon. Cinematic, atmospheric, sensory.
Result: Neon reflection on wet pavement was one of the most faithful outcomes from any city-night prompt we've run. Long-exposure light trails on cars came through cleanly. Pedestrian blur was correct for long exposure rather than ghosting. Strong for a travel, automotive, or luxury brand.
Variant: Add a single static figure standing in the rain mid-frame to anchor the shot.
Veo 3.1 Specific Techniques
Dialogue Prompts
Wrap spoken lines in quotation marks and specify the speaker. "She looks at the camera and says, 'Come find us.'" works. Adding "lip sync, audio included" as an instruction reduces the rate of mouth-movement drift. Budget two to three takes for dialogue prompts. The model nails it about 60% of the time on the first generation.
Audio Cues
Veo 3.1 generates synchronized audio alongside video. You can specify ambient sound type ("rain on glass, distant traffic, no music"), score mood ("sparse piano, minor key, unresolved"), or sound effects ("high heels on marble, echo"). Include audio instructions in the mood section at the end of the prompt for best results.
Motion Blur Control
Specify shutter equivalent to control motion blur. "1/48 shutter equivalent, 24fps" gives you classic film motion blur. "1/250 shutter equivalent" gives you crisp motion for product shots. "1/60 shutter equivalent, 60fps" gives you the hyperreal sports look. This is one of the most underused Veo levers.
Camera Move Language
Veo 3.1 responds well to these specific move names:
- Push-in / Pull-out: dolly move toward or away from subject
- Tracking shot: camera moves parallel to a moving subject
- Crane up / Crane down: vertical camera rise or fall
- Dutch angle: camera tilted on its axis, specify degrees ("15-degree Dutch angle")
- Rack focus: "rack focus from foreground object to subject" or reverse
- Whip pan: fast horizontal pan, specify direction
Compound moves work: "start static, slow push-in beginning at 2 seconds, ending on extreme close-up." Veo 3.1 will distribute the move across the clip.
Common Veo 3.1 Prompt Failures
Muddy lighting: happens when you specify multiple conflicting light sources without directions. "Bright sunlight and candlelight and neon signs" with no spatial information. Fix: give each source a direction. "Candlelight from screen left, neon from the window behind, no fill."
The flicker artifact: appears on reflective surfaces (glass, water, metal) when you underspecify the environment. The model generates a light source but can't decide where it is, so it moves it. Fix: anchor your key light to a specific source and direction. "Hard key light from a single overhead fixture, 6 feet up, camera right."
Physics failure (cloth, liquid, hair): usually caused by a contradiction between the described motion and the lighting conditions. Fast motion in still lighting, or still subject in a "windy" environment, confuses the physics solver. Be consistent: if the subject is moving fast, the environment should reflect that. If it's still, commit to stillness.
Character inconsistency mid-clip: Veo doesn't have reference conditioning. If you describe your character in the first line, the model holds the description for about 3 to 4 seconds before it starts to drift. For clips longer than 5 seconds, keep the character description prominent and specific. Avoid long environment descriptions that push the character description far from the start.
Killed physics: the phrase "perfect symmetry" or "perfectly still" applied to natural subjects (water, cloth, hair, smoke) tends to suppress the physics engine. Drop that language and let the model handle subtle natural motion.
Combining Veo with Reference Images
Veo 3.1 supports single reference image input for visual style conditioning. It's not the same as Seedance's multi-reference product conditioning (covered in our best AI video generator comparison), but it's useful for matching a brand's visual grammar.
On 8frame, you pass the reference image through the canvas before the Veo generation step. The model uses it to calibrate color temperature, contrast, and stylistic language, not for subject identity. A reference image of a fashion editorial gives you the color palette and lighting quality. A product shot gives you the surface and light style. Don't expect it to lock character appearance: for that, Seedance or Higgsfield are better choices.
A typical 8frame chain for brand film work: generate or source a reference still, pass it through Veo 3.1 with your text prompt, then upscale the output. See 8frame workflows for the template.
FAQ
How long should a Veo 3.1 prompt be?
60 to 120 words is the working range. Shorter than 60 words leaves too many elements for the model to fill in by default. Longer than 120 words and the model starts to lose track of elements it encountered early in the prompt. The six-element formula (Subject + Action + Environment + Camera + Lighting + Mood) naturally lands in this range.
Does Veo 3.1 support dialogue?
Yes. Wrap the spoken line in quotation marks and specify the speaker in the subject element. Dialogue prompts work on roughly 60% of first-generation attempts. Run two to three takes and use the cleanest. The audio and lip movement are generated together, so you don't have to do separate audio work for short dialogue cuts.
Can Veo 3.1 handle multiple characters?
Two characters work well. Three is where prompt-adherence starts to degrade. For two characters, describe both fully at the start of the prompt, specify their relative positions ("one standing, one seated across from her"), and keep the action clearly attributed to one or the other. For three or more characters, consider using Higgsfield Soul 2.0, which handles group shots and character consistency better at this count.
What aspect ratio is best for Veo 3.1?
16:9 for cinematic and brand film work. 9:16 for social-first deliverables. Veo 3.1 handles both natively. The model's default is 16:9. Specify 9:16 in the environment or camera element ("vertical 9:16 frame, social crop") and it will compose the shot for that format rather than cropping a 16:9 output. Avoid 1:1 for Veo 3.1: the composition logic is weaker in square format.
What changed between Veo 3 and Veo 3.1 for prompting?
The main changes are better dialogue sync, improved motion blur handling, and more reliable camera-move execution. Veo 3 would sometimes flatten a specified camera move into a subtle zoom. Veo 3.1 executes named moves more faithfully. Lighting language also got more responsive: you can specify stops of exposure in 3.1 ("two stops underexposed") and get a meaningful result. Prompts written for Veo 3 work in 3.1 without changes, but you can push further with the additional controls 3.1 responds to. See our Veo 3 vs Sora 2 vs Kling 3 comparison for a full breakdown of model generations.
Run These Prompts on 8frame
Every prompt in this guide was generated on the 8frame canvas using Veo 3.1. If you want to test any of them without building the setup from scratch, the workflow template is open. Drop in your subject, swap the environment, and iterate from a prompt that already has the formula right.