Prompt guide
VEO 3.1 prompt guide
VEO 3.1 writes its own soundtrack, so a good VEO prompt describes what you hear as carefully as what you see: put the spoken line, the sound effects and the room tone in their own sentences after the picture. On Viralance a 4-second VEO 3.1 Lite test costs 3 credits ($0.90) and an 8-second VEO 3.1 Fast clip 12; Fast finished 98% of 149 real user jobs, the highest rate in our generation data.
Updated November 7, 2026. Viralance is an independent platform that runs these models; it is not affiliated with their makers.
Versions this guide covers
| Version | Length | Resolution | Sound | Start + end frame | Credits |
|---|---|---|---|---|---|
| VEO 3.1 Fast | 4 / 6 / 8s | 720p / 1080p | Yes | Yes | 6–12 |
| VEO 3.1 Lite | 4 / 6 / 8s | 720p / 1080p | Yes | Yes | 3–8 |
The prompt structure
Shot and lens → subject → action → setting and light → style → then, in separate sentences: dialogue, sound effects, ambience
- Shot and lens
- Start with the framing: "Close-up", "Wide shot", "Handheld selfie-style video", "Macro shot". Google lists camera angle, movement and lens effects as separate prompt elements; VEO weights what comes first.
- Subject
- One main subject with two or three visible details (age range, clothing, material). Describe a fictional person, never a named one; named people are blocked before the job starts.
- Action
- One action that fits the length. An 8-second clip holds one gesture and one line of speech comfortably; three actions get rushed or dropped.
- Setting and light
- Place and time of day in a few words: "bright kitchen, morning window light", "neon-lit street after rain".
- Style
- A look, not a list of adjectives: "70s film look", "clean commercial lighting", "documentary handheld".
- Sound, written last
- Google's guide says to write audio in separate sentences. Give the spoken line after "says:", then the effects ("glass clinks"), then the ambience ("quiet cafe chatter"). If you want silence, switch Sound off in the studio instead of writing "no audio".
Medium close-up, handheld. A woman in her 30s with a green apron stands behind a bakery counter and slides a tray of croissants toward the camera. Warm morning light through the shop window, flour dust in the air, natural colors. She says: "First batch of the day, still warm." The tray scrapes softly on the counter. Quiet street noise outside.
Writing dialogue VEO can say
VEO 3.1 generates speech and lip movement together. Keep each line short enough to say in the clip: about 15 to 20 words fit in 8 seconds, half that in 4. A longer line gets cut off mid-word or spoken too fast.
Attach the line to the speaker: "The older man says:" works better than a floating quotation when two people are in the shot. For a two-person exchange in 8 seconds, give each person one short line.
Write the delivery next to the line if it matters ("whispers", "laughs and says"). Accents and languages are less predictable; test them on Lite at 4 seconds first.
Sound effects and ambience
Name sounds as events the viewer would notice: "the lid pops", "a dog barks twice in the distance", "rain on the car roof". Ambience is the background bed: "office hum", "crowded market", "wind through pines".
VEO also adds music when the scene suggests it. If you plan to put your own track or a voiceover on top, say "no music" in the sound sentence or keep the ambience simple, then mix later in Audio Suite.
Audio can trigger a refusal on its own: when the provider flags the generated soundtrack, Viralance resubmits the same job once with sound off before marking it failed, and a failed job is refunded. If a prompt keeps failing that way, rewrite the sound sentence (no song titles or lyrics) or turn Sound off.
Fast or Lite for a given prompt
Lite is the drafting model: 3 credits for 4 seconds and 5 for 8 seconds at 720p, with sound. Fast costs 12 credits for 8 seconds at either 720p or 1080p and supports Reference mode with up to 3 images.
In our data from 2026-01-29 to 2026-11-09, Fast completed 98% of 149 jobs and Lite 92% of 1002, with a median wait of 2.3 and 2 minutes. Most Lite failures were policy refusals, and they reflect what people asked for more than how strict the model is.
A practical loop: write the prompt, run it on Lite at 4 seconds, fix what came out wrong, then run the final on Fast at 8 seconds.
- VEO 3.1 Fast image-to-video runs at 8 seconds in the studio (12 credits). Lite animates a photo at 4, 6 or 8 seconds.
- In image-to-video mode the clip keeps the photo's shape; in text-to-video you pick 9:16 or 16:9 under Format.
- The studio limit is 2,000 characters per prompt. VEO rarely needs more than 80 words.
Camera words VEO 3.1 follows
| Write | Use it for |
|---|---|
| slow dolly in | Product reveals and emotional beats; the background stays steady while the subject grows. |
| handheld, selfie-style | UGC ads and talking clips; adds small natural shake and arm-length framing. |
| drone shot pulling back | Landscape and travel openers that end on a wide reveal. |
| tracking shot following | A person walking or a car driving; the camera keeps pace beside or behind. |
| static shot, tripod | Dialogue and demos where any drift would distract. |
| macro close-up | Food, cosmetics, textures, droplets; pair with "shallow depth of field". |
| low angle | Makes a subject feel large: sneakers, cars, a hero shot. |
| over-the-shoulder | Two-person scenes and screen reveals. |
| rack focus from … to … | Moving attention between two objects in one take. |
| whip pan | Energetic transitions; works best at the start or end of a clip. |
Start and end frames
Both VEO 3.1 models take a start and an end frame. Choose Image to video, upload the first frame, then add the optional End frame; the clip moves from one picture to the other. It costs the same as plain image-to-video: 12 credits for 8 seconds on Fast, 4 for 6 seconds on Lite at 720p.
The prompt should describe the change between the two pictures, not the pictures themselves: "the camera slowly pushes in as the lights in the room switch on" for a dark and a lit version of the same room.
- Keep framing, lens and lighting direction close between the two frames; VEO has to invent everything that differs.
- Same subject, same scale: a full-body start and a face close-up end produces a warped zoom.
- Use Swap in the studio to reverse a transformation without uploading again.
- For more than two key moments, Kling 4.0 takes up to 8 keyframes between start and end.
Refused or misread prompts, and the fix
A named celebrity or a "make [person] say" prompt
Viralance blocks these before the job starts and nothing is charged. Describe a fictional person by age, look and clothing instead. More in why AI video prompts get rejected.
A photo of a real face comes back with no video
VEO declines some photo inputs by returning nothing; 4 of the Lite photo jobs in our data ended that way. Use a photo of yourself or someone who agreed, tick the consent box, or start from an AI-generated character. See why image-to-video fails.
The soundtrack is flagged
Song titles, lyrics and famous jingles invite this. Describe the mood of the music instead ("upbeat acoustic guitar"), or switch Sound off and add music afterwards.
Brand names and logos
Describe the object without the trademark ("matte black over-ear headphones"). For your own product, start from your product photo in image-to-video mode.
"No people" or "don't show text" is ignored
Google advises against instructive negatives like "no" and "don't". Describe what should fill the frame ("an empty beach at dawn") so the unwanted thing has no place in the scene.
10 VEO 3.1 prompts to copy
Each one opens in the studio with the model, mode and prompt filled in. Photo prompts need your own image uploaded first.
- VEO 3.1 Fast · text · 8s · 720p · 12 credits
1. UGC skincare testimonial
Handheld selfie-style video in a bright bathroom. A woman in her late 20s with curly hair holds a small white jar of face cream up to the camera and smiles. Natural morning light, no makeup look. She says: "Two weeks in and my skin finally stopped flaking." Soft bathroom echo.
One line of about ten words leaves room for the smile at the end.
Use this prompt - VEO 3.1 Fast · text · 8s · 1080p · 12 credits
2. Two-person diner scene
Medium shot of two friends in a 1970s diner booth. The man knocks over a milkshake; the woman leans back laughing. Warm tungsten light, grainy film look. The woman says: "You did not just do that." The man says: "Not a word." Diner clatter and a jukebox far away.
Each speaker is named before the line; two short lines fit 8 seconds.
Use this prompt - VEO 3.1 Lite · text · 6s · 720p · 4 credits
3. Coffee pour, sound-led
Macro close-up of espresso pouring into a clear glass of iced milk, slow motion, the coffee clouds into the milk. Dark kitchen counter, a single window light from the left. Ice cracks as the hot coffee hits it. Quiet morning room tone.
No dialogue; the ice crack is the hook. Cheap enough to run three variants.
Use this prompt - VEO 3.1 Lite · text · 8s · 1080p · 8 credits
4. Coastal drive opener
Drone shot pulling back from a red vintage convertible driving along a cliff road at golden hour, revealing the ocean and the coastline. Cinematic, warm tones. Engine purr, waves below, wind.
1080p on Lite is a low-cost way to get a wide establishing shot.
Use this prompt - VEO 3.1 Fast · photo · 8s · 1080p · 12 credits
5. Product reveal from your photo
Slow dolly in toward the bottle on the marble counter. Soft light sweeps across the glass from left to right and a few water droplets roll down the side. The label stays sharp and unchanged. A gentle chime as the light passes. No music.
Upload a clean product photo; the prompt only adds motion and light, so the label is less likely to change.
Use this prompt - VEO 3.1 Lite · text · 6s · 720p · 4 credits
6. Street food vendor
Close-up, handheld. A street food cook in a night market flips noodles in a smoking wok, flames jump up for a second. Neon signs blur behind him. Sizzling oil, the clang of the spatula, crowd chatter.
Food scenes carry well on sound alone: list two or three distinct noises.
Use this prompt - VEO 3.1 Fast · text · 6s · 720p · 8 credits
7. Explainer to camera
Static shot, eye level. A man in his 40s in a grey sweater sits at a tidy desk with a laptop and talks to the camera, calm and friendly. Soft office light, plants in the background. He says: "Here is the one setting most people never change." Quiet room tone.
A static camera keeps the face steady for a talking hook; cut it before your real screen recording.
Use this prompt - VEO 3.1 Lite · text · 8s · 720p · 5 credits
8. Rainy window mood loop
Macro shot of rain running down a window at night, city lights blurred into bokeh behind the glass, a cat silhouette on the sill slowly turns its head. Soft steady rain, distant traffic.
Little motion and a steady sound bed make a clean loop for a background or a lyric video.
Use this prompt - VEO 3.1 Lite · photo · 8s · 720p · 5 credits
9. Day to night, start and end frame
The camera holds still as the sun sets: the sky shifts from blue to deep orange to dusk, the windows of the house light up one by one, and the garden lamps switch on. Evening birdsong fading into crickets.
Upload a daytime photo as the start frame and an evening photo of the same view as the end frame.
Use this prompt - VEO 3.1 Lite · text · 6s · 720p · 4 credits
10. Pet reaction clip
Low angle, handheld. A golden retriever puppy sits on a kitchen floor and tilts its head at a squeaky toy held just off camera, then pounces on it. Bright natural light. Two squeaks, puppy paws sliding on tiles, a person laughing off camera.
An off-camera laugh adds life without putting a person in the frame.
Use this prompt
Running a VEO 3.1 prompt on Viralance
- 1
Draft the picture in one sentence
Write the shot, subject, action and setting first, in that order, in under 40 words.
- 2
Add sound as separate sentences
After the picture, write the spoken line after "says:", then one or two sound effects, then the ambience. Switch Sound off instead if you will add your own audio.
- 3
Test on VEO 3.1 Lite
Open the studio with VEO 3.1 Lite, choose 4 seconds at 720p and generate. The cost shows before you start.
- 4
Fix one thing at a time
Change only the part that came out wrong (camera, line length, sound) so you know what fixed it.
- 5
Render the final on Fast
Switch to VEO 3.1 Fast at 8 seconds and 1080p for the clip you will publish, or add an end frame in image-to-video mode.
VEO 3.1 prompt questions
How long should a VEO 3.1 prompt be?
Most good VEO prompts are 40 to 90 words: one sentence for the shot, one or two for the action and setting, and separate sentences for dialogue and sound. The Viralance studio accepts up to 2,000 characters, but extra detail beyond that range usually gets ignored or crowds the action.
How do I make a character speak in VEO 3.1?
Name the speaker and give the line after "says:", for example: The woman says: "Still warm." Keep it to about 15 to 20 words for 8 seconds. Sound must be on, which it is by default in the studio.
Can I use negative prompts with VEO 3.1?
The studio has no separate negative prompt field for VEO. Google advises against instructive phrasing such as "no" or "don't" in the prompt; describe what should be in the frame instead. To remove all sound, switch Sound off.
Does VEO 3.1 support first and last frames?
Yes. In image-to-video mode, upload a start frame and add an end frame; both VEO 3.1 Fast and Lite move from one to the other. It costs the same as image-to-video, for example 12 credits for an 8-second Fast clip.
Why did my VEO prompt get refused?
The usual causes are a real or famous person, a photo with a real face, sexual or violent content, trademarks, or a soundtrack flagged as copyrighted. Provider refusals are refunded automatically. Rewrite the request without those elements and change one thing per retry.