Prompt guide · Veo 3.1
Veo 3.1 Prompt Guide
Veo 3.1 rewards people who write prompts like shot lists, not wish lists. It executes camera vocabulary literally, speaks dialogue exactly as quoted, and treats sound as part of the scene. This guide covers the structure that works, with prompts you can run as-is — every example opens directly in the Veo 3.1 editor.
Last updated July 11, 2026
01Structure: subject → action → camera → light → sound
Veo parses prompts best when they read like a director's brief in one breath: who or what we're looking at, what happens, how the camera behaves, what the light does, and what we hear. Order matters less than completeness — clips fail most often because a prompt never says what the camera is doing.
02Camera moves it takes literally
Orbit, dolly, crane, pan, tilt, rack focus, handheld, FPV — Veo 3.1 executes these as named. Chain at most two moves per clip; three or more starts to blur into mush. If you don't specify a move, you'll get a slow generic push-in, which reads as filler.
“Low-angle crane up the face of a brutalist apartment block at dusk, laundry lines crossing between balconies, then rack focus to a cat on the nearest railing, city hum below”
Run this prompt →“FPV drone shot threading through a narrow slot canyon at golden hour, sandstone walls inches from the lens, wind and echoing rotor buzz”
Run this prompt →03Dialogue: quote it, and it gets spoken
Put spoken lines in quotation marks and describe the speaker first. Veo 3.1 will lip-sync the exact words — this is its biggest edge over other models, so use it for hooks and to-camera moments. Keep lines under ~15 words per 8-second clip; longer speeches get rushed.
04Direct the sound, not just the picture
Audio is generated with the video, so treat it as a first-class element: name the ambience, the effects, and the silence. "No music" is a valid and useful instruction — without it, Veo sometimes adds a generic score that fights your edit.
05What wastes credits
The recurring failure modes: prompts with two competing subjects (pick one), abstract adjectives with no physical referent ("epic", "stunning"), requesting on-screen text (use an image model for title cards instead), and cramming a story arc into 8 seconds. One subject, one move, one beat — then extend or cut together in the video tools.
Frequently asked questions
How long should a Veo 3.1 prompt be?
Two to four sentences covering subject, action, camera, light, and sound. Shorter prompts under-specify the camera; longer ones bury the instruction that matters.
Can Veo 3.1 speak exact dialogue?
Yes — put the line in quotation marks and describe the speaker. The words are spoken as written and lip-synced.
How do I stop Veo from adding music?
Say it: end the prompt with "no music" and name the sounds you do want. Audio direction is followed as literally as camera direction.
Do these prompts work for image-to-video too?
Yes, with one change: drop the subject description (the image supplies it) and spend the prompt on motion, camera, and sound.
More guides
Put the guide to work
Every example above opens in the Veo 3.1 editor with the prompt loaded — free credits on signup.
Open Veo 3.1