Veo 3.1 Prompts for Narrated Story Scenes (No Dialogue)
A step-by-step way to write Veo 3.1 prompts for short story videos told by a narrator: scenes with ambient sound but no talking, a consistent main character, and shots that cut together cleanly.
Veo 3.1 is known for clips where characters talk. Give it a line in quotes and it voices it, with lip movement and background sound in the same pass. That is great for skits. It gets in the way when you are making a narrated story, where one voice tells the tale and the characters on screen simply live it.
This tutorial shows how to write Veo 3.1 prompts for that second kind of video: short story scenes with no dialogue, ambient sound only, and a main character who looks the same from shot to shot. It works with the Flow AI video generator on this site, where Veo 3.1 makes 4, 6 or 8-second clips in 16:9 or 9:16.
Step 1: Write the narration before any prompt
The narration is the spine of the video, so write it first and read it aloud with a timer. Then split it into chunks of one sentence or two, each roughly the length of one clip. A 4-second clip fits a short sentence; an 8-second clip fits two.
Each chunk becomes one shot. Write next to it what the viewer should see, in plain words:
| Narration | What we see |
|---|---|
| "Every night at three, Mara unlocked the bakery." | Wide shot, dark street, one lit window |
| "Nobody in town had ever seen her bake." | Mara kneading dough alone, flour in the air |
If you would rather hand the whole job to one tool, ViewMax Studio takes a different route: its ViewMax Agent 2.0 plans its own shot list from your brief, renders the scenes, reads a narration over them and times on-screen text, all in one render. This tutorial is for doing it shot by shot, with full control over every scene.
Step 2: Use one scene template for every shot
Write every prompt in the same order so the shots feel like they belong together:
- Shot type: wide, medium, close-up.
- The character line: the same words in every prompt (see step 3).
- One action.
- Setting and light.
- Camera move.
- Audio line: what you hear, and what you don't.
The audio line is the important one for narrated stories. Veo 3.1 adds sound to every clip, so say what that sound should be and rule out the rest: "Ambient sound only: rain on the window, a ticking clock. No dialogue, no one speaks, no music." Leave out anything in quotation marks, since quoted text is how you ask Veo for spoken lines.
Step 3: Lock the main character
Characters drift between separate clips because each prompt is a fresh generation. Two habits help.
A fixed character line. Write one sentence describing the character and paste it unchanged into every prompt: "Mara, a woman in her sixties with short grey hair, round glasses and a flour-dusted green apron." Don't paraphrase it between shots.
A picture as the first frame. Better still, make a still of the character first with the Flow AI image generator, then start each shot from an image with Flow AI image to video. When the first frame already shows the character, Veo only has to animate her, not reinvent her.
Step 4: Six prompts for a one-minute story
Here is a complete short, "The Night Baker", as six shots. Each one uses the template above.
Shot 1, 8 s, wide
Wide shot of a quiet small-town street at 3 a.m., wet cobblestones, one bakery window glowing warm yellow. Mara, a woman in her sixties with short grey hair, round glasses and a flour-dusted green apron, unlocks the door and steps inside. Slow push-in toward the window. Ambient sound only: distant wind, a key turning. No dialogue, no music.
Shot 2, 6 s, medium
Medium shot inside a small bakery. Mara, a woman in her sixties with short grey hair, round glasses and a flour-dusted green apron, kneads dough on a wooden table, flour drifting in the lamp light. Static camera. Ambient sound only: dough slapping the table, a clock ticking. No one speaks, no music.
Shot 3, 4 s, close-up
Close-up of Mara's hands shaping a round loaf and setting it on a tray. Shallow depth of field. Ambient sound only: soft thuds of dough. No dialogue.
Shot 4, 8 s, medium
Medium shot. Mara, a woman in her sixties with short grey hair, round glasses and a flour-dusted green apron, opens the oven, warm orange light fills her face and glasses, she slides the tray in and closes the door. Slow dolly to the right. Ambient sound only: oven door creak, crackle of heat. No dialogue, no music.
Shot 5, 6 s, wide
Dawn light over the same street. Mara, a woman in her sixties with short grey hair, round glasses and a flour-dusted green apron, places a basket of fresh bread on the doorstep of a neighbour's house and walks away. Static wide shot. Ambient sound only: early birds, footsteps on stone. No one speaks.
Shot 6, 8 s, wide
The bakery window from outside at sunrise. Mara, a woman in her sixties with short grey hair, round glasses and a flour-dusted green apron, visible inside, turns off the lamp, the street slowly filling with morning light. Slow pull-back. Ambient sound only: birds, a distant bicycle bell. No dialogue, no music.
Notice that faces are often turned away, in profile, or small in the frame. That is deliberate: when no one speaks, there are no mouths to animate, and the narrator's voice never fights a moving pair of lips.
Step 5: Keep shots flowing into each other
Hard cuts between unrelated clips read as a slideshow. Two techniques make them feel continuous:
- Match the frame. Take the last frame of a clip as the first frame of the next, so the light and position carry on.
- Use first and last frame. Switch the box to first and last frame mode and give Veo both ends of the shot, for example Mara at the table and Mara at the oven, and it works out the movement in between.
This last-frame-to-first-frame chaining is what a long video tool that films shot after shot automates: it plans a shot list, renders each MiniMax H3 shot from the end of the one before, and joins them into one video of up to 4 minutes. Doing it by hand with Veo 3.1 gives you more control over each shot; automating it saves the frame-grabbing.
Step 6: Assemble in your editor
- Lay down the narration track first.
- Drop each clip under its sentence, trimming to the end of the line.
- Turn Veo's ambient sound down so it sits under the voice, not over it.
- Add music in the editor, where you control the volume, rather than asking Veo for it.
- Add captions last.
When a character still talks
Sometimes a clip comes back with a murmur or a mouthed word. Remove anything that sounds like a line, add "no one speaks" as well as "no dialogue", keep the face turned or small, and regenerate. If the picture is perfect and only the sound is wrong, keep the clip and mute it in the editor; the narration covers it.
When to skip shot-by-shot
Six shots by hand is a pleasant evening. Sixty is not. For a story that runs several minutes, a dedicated story tool is the faster route: ToonBee AI turns a story or a script into a narrated video of up to 15 minutes, with the same characters drawn in every scene. Its story videos use one narrator and characters do not speak lines to each other, which is the same format this tutorial builds by hand.
FAQ
Can Veo 3.1 make a clip with no speech?
Yes, describe the sound you want and add "no dialogue, no one speaks". Veo 3.1 always returns audio, so if a stray voice appears, regenerate or mute that clip in your editor.
How long can each scene be?
Here, Veo 3.1 Lite and Veo 3.1 Fast make 4, 6 or 8-second takes. For a longer scene, join several takes.
Which Veo 3.1 tier should I use for story scenes?
Veo 3.1 Lite renders 720p at the lowest price and is fine for drafting every shot. Veo 3.1 Fast adds finer detail and 1080p, so it suits the final version of your key shots.