zmime.com
Video model · Sora

Prompt for Sora

Write Sora prompts like shot cards. One subject, one action, one camera move, one light setup.

  • Text to video
  • Camera language
  • Continuity
  • Short clips
01 — Overview

What Sora rewards

Sora generates short clips, so the unit of work is a shot, not a scene. Prompts that describe a sequence of events tend to produce drift, warped subjects or unrequested cuts. Prompts that describe a single continuous action with a single camera movement hold together much better.

Borrow vocabulary from film production: shot size, lens, camera move and speed, key light direction, grade. It is the most precise language available for what you actually want, and these models were trained on plenty of it.

  • 01Coherent physical motion within a single shot.
  • 02Responds well to cinematography terminology.
  • 03Useful for b-roll, product motion and atmosphere.
  • 04Weak on readable text, precise brand geometry and long dialogue.
02 — Anatomy

The Sora shot card

  • Shot size Wide, medium, close-up, macro. Establishes scale first.
  • Subject One subject with material and wardrobe detail.
  • Action One continuous action that finishes inside the clip length.
  • Camera Move plus speed, or "static, locked off".
  • Light Key source, direction, quality, and that it stays constant.
  • Lens + grade Focal length, aperture, colour treatment, grain.
  • Exclusions No cuts, no text overlays, no additional subjects.
03 — The prompts

4 prompts written for Sora

Paste, then replace the [bracketed] parts. Keep the wording identical for anything you want repeated across a set.

Nothing matches that word. Clear the field to see every prompt.

Prompt 01

Single hero shot

The main clip in a piece, with controlled motion.

  • hero
  • camera
Template · plain text 
Medium shot, 6 seconds, one continuous take.

SUBJECT: [subject], [wardrobe/material], [state or expression].
ACTION: [one action that begins and completes within 6 seconds].
CAMERA: slow push in, roughly 20cm over the clip, steady on a dolly, no shake, no cuts.
ENVIRONMENT: [location] at [time of day], [two background details].
LIGHT: [source] from camera-left, soft shadows, brightness constant throughout.
LENS: 40mm, f/2.8, subject sharp, background gently soft.
GRADE: natural colour, mild filmic contrast, fine grain.
AUDIO: [ambience only].

No cuts, no text or captions, no additional people entering frame.
Prompt 02

Atmospheric establishing shot

Opening frames that set place and mood.

  • establishing
  • atmosphere
Template · plain text 
Wide establishing shot, 5 seconds.

SUBJECT: [location], with [subject] small in frame for scale.
MOTION IN SCENE: [environmental movement only — drifting fog, moving water, blowing fabric].
CAMERA: static, locked off on a tripod. No movement at all.
TIME AND WEATHER: [time of day], [weather], visible atmosphere with depth layers.
LIGHT: [natural source] low in frame, long shadows, consistent exposure.
LENS: 24mm, f/5.6, deep focus.
GRADE: [cool shadows, warm highlights], gentle halation, subtle grain.

Nothing enters or leaves frame. No camera move, no cuts, no text.
Prompt 03

Product motion clip

Loopable website or ad footage.

  • product
  • loop
Template · plain text 
Close-up, 5 seconds, designed to loop seamlessly (first and last frame match).

SUBJECT: [product], [material and finish], [colour], resting on [surface], perfectly still.
ACTION: none — the product does not move.
CAMERA: smooth 45° orbit clockwise at constant speed, fixed height, no acceleration or easing.
LIGHT: soft top light producing a specular highlight that travels across the surface as the camera moves, subtle rim light behind.
BACKGROUND: [flat colour], clean, out of focus.
LENS: 100mm macro, f/4, focus locked on [key detail].
GRADE: neutral, accurate product colour.

No hands, no text, no reflections of a studio, no objects entering frame.
Prompt 04

Continuity block for a multi-shot sequence

Keeping character and place consistent across generations.

  • continuity
  • sequence
Template · plain text 
I am generating [n] separate clips that must cut together. Repeat this block verbatim at the top of every prompt, changing only the SHOT section.

CONTINUITY (identical in every clip)
- Character: [fixed description — age, build, hair, distinguishing feature]
- Wardrobe: [fixed garments and colours]
- Location: [fixed location], [fixed time of day], [fixed weather]
- Light: [fixed key source and direction]
- Lens family and grade: [fixed], consistent colour treatment

SHOT (changes per clip)
- Shot size: [wide / medium / close]
- Action: [one action, completes within the duration]
- Camera: [one move with speed, or static]
- Duration: [n] seconds

RULES FOR EVERY CLIP
- One continuous take, no cuts, no text overlays.
- Match screen direction: [subject moves left to right] throughout.
- Do not introduce new characters, props or locations.
04 — Craft notes

What separates a good prompt from a wasted run

Do this 05

  • Describe one shot per prompt and edit clips together yourself.
  • State camera movement with a speed, or say static.
  • Match action length to clip length.
  • Repeat a continuity block verbatim across a sequence.
  • Generate multiple takes and pick, rather than re-prompting blindly.

Not this 05

  • Multi-scene narratives in one prompt.
  • Two simultaneous camera moves.
  • Long dialogue in a short clip.
  • Expecting accurate logos or on-screen text.
  • Vague energy words instead of motion description.
05 — Questions

Asked often, answered plainly

Why do subjects warp mid-clip?

Too much motion for the duration, or competing motions. Reduce to one moving element, slow the camera, or shorten the action so it can finish cleanly.

How do I get consistent characters across clips?

Reuse a fixed continuity block word for word, keep wardrobe and lighting identical, and start from a reference frame where the tool allows it. Any rewording reads as a new person.

Can I generate a full 30-second ad in one go?

Not reliably. Plan it as six to eight shots, generate each separately with a shared continuity block, then assemble in an editor. You also get to reject individual shots instead of the whole thing.

06 — Sources

Read the primary documentation

Vendors change flags, limits and defaults often. Where this site disagrees with official docs, the docs win.

07 — Categories

Browse by output instead

  1. 01

    Video prompts

    One shot, one movement, one action. Video prompts fail when they try to be a whole scene.

    • Text to video
    • Camera moves
    • B-roll
     5 prompts
  2. 02

    Avatar prompts

    The hard part of avatars is not the first image, it is the second one looking like the same person.

    • Profile pictures
    • Character sheets
    • Mascots
     5 prompts