Why Prompts Matter

Sora 2 aims to understand scenes and physics via world modeling. Strong prompts help the model maintain character, scene, and camera consistency over longer durations—and make motion, lighting, and texture more controllable.

This guide distills our on‑site experience into templates and examples that are reproducible, controllable, and easy to iterate.

Ground Rules (for both Text‑to‑Video and Image‑to‑Video)

• State camera intent: describe with cinematography, not adjectives alone
• Avoid contradictions: don’t mix opposite styles/lighting/pacing
• One focus at a time: split complex goals into multiple shots/rounds
• Short first, longer later: validate direction with a minimal prompt, then add detail
• Fix key factors: keep subject, style, and lighting stable; tweak one variable at a time
• Use references: for image‑to‑video, upload a high‑res subject; use text for camera and actions

Universal Template (Text‑to‑Video)

Scene: place/time/mood (e.g., rainy city at night, puddle reflections)
Subject: who/what/state (e.g., woman in trench coat walking right‑to‑left)
Narrative/Action: clear actions and interactions
Camera: moves (dolly/zoom/pan/tilt/follow), focal length (35mm/85mm), depth (shallow/deep)
Lighting: key/fill/ambient (back/side/top), reflections/volumetric light
Texture/Style: photoreal/film/illustration/cyberpunk/commercial grading
Composition: symmetry/rule of thirds/low angle/centered; foreground/mid/background
Pacing/Duration: slow push/steady/fast; define the ending (e.g., freeze frame)
Aspect: 16:9 / 9:16 / 1:1

Example (cinematic realistic):

A rainy city at night, puddle reflections; the subject is a woman in a trench coat walking right‑to‑left;
Handheld, 35mm look, shallow depth of field, slow push in to a close‑up;
Backlight with neon reflections and slight haze; realistic style, cool cinematic grading;
Rule‑of‑thirds composition, foreground lamppost passes by; slow pace, end on a freeze frame; 16:9.

Universal Template (Image‑to‑Video)

Use the uploaded image as reference; keep the subject’s appearance consistent;
Camera script: push from medium shot to close‑up (or orbit/track), keep the subject sharp;
Background dynamics: slight dynamic bokeh/environment motion for parallax and depth;
Lighting/style: keep original light direction and color tone (e.g., clean cool commercial look),
  moderately enhance highlights/reflections;
Pacing/Ending: steady pacing, end on a subject close‑up freeze;
Aspect: 9:16 (vertical short).

Example (product visual):

Preserve the product’s look and materials; camera pushes from medium to close‑up;
Emphasize metallic edge highlights and surface reflections; background with slight dynamic bokeh;
Clean commercial style with cool tones; steady pace; end on the logo close‑up; 9:16.

Sora 2 Prompt Keyword Bank (copy and combine)

• Camera movement: dolly, zoom, pan, tilt, truck, orbit, crane, handheld, steadicam, low angle, high angle, overhead, long take
• Focal length / DoF: 24mm, 35mm, 50mm, 85mm, wide, telephoto, shallow depth of field, deep focus, bokeh, dynamic bokeh
• Composition: rule of thirds, centered symmetry, foreground occlusion, leading lines, low/high angle, negative space, golden ratio
• Lighting / materials: backlight, sidelight, toplight, rim light, volumetric light, hard/soft light, neon reflections, metallic highlights, glass refraction, skin specular
• Style / texture: photoreal, film grain, cinematic grading, clean commercial, vintage, cyberpunk, illustration, anime, film noir, minimal
• Motion / physics: slow motion, time‑lapse, wind/rain/snow/fog, dust/particles, cloth simulation, fluids, splash, reflection/refraction
• Pacing / transitions: slow push, steady, fast cuts, montage, jump cut, fade in/out, dissolve, end on freeze frame
• Aspect: 16:9, 9:16, 1:1, 2.39:1 (widescreen)
• Color / grading: cool/warm tone, complementary palette, teal‑orange, low saturation, high contrast, natural skin tones
• Rig vibe: handheld, gimbal, slider, jib, drone, top‑down/overhead
• Character consistency: hairstyle, outfit, primary color, props (e.g., red scarf, black leather gloves)

Example combo:

Handheld steady, 35mm shallow DoF, low‑angle slow push; backlight + neon reflections, rule‑of‑thirds; cool cinematic grade; 16:9.

FAQ: Sora 2 prompts

• What’s different about a “Sora 2 prompt”?
  Use cinematography + physics cues (e.g., slow push, backlight, volumetrics). Sora 2 rewards scene/physics‑aware phrasing with better consistency.
• Any ready‑to‑use Sora 2 prompt keywords?
  See the “Sora 2 Prompt Keyword Bank” above. Copy a base combo, then adapt to your project’s style.
• Ideal length for a Sora 2 prompt?
  Short first, longer later: validate with 1–2 lines, then add camera/lighting/composition/style. Avoid piling up contradictory adjectives.
• How do text‑to‑video vs image‑to‑video prompts differ?
  Image‑to‑video prioritizes “subject consistency” from the image; use text to script camera and motion. Text‑to‑video must also define subject/lighting/style.

Want broader context and examples? Read our intro: What Is Sora 2.

Advanced Control & Common Issues

• Character consistency: fix wardrobe/hair/tone keywords across iterations
• Motion intensity: be explicit (slow push/steady/small pan) to avoid jitter
• Depth and focal length: specify shallow DoF / 35mm / 85mm to reduce edge blur/pop
• Lighting/reflections: call out backlight/side light/neon reflections/volumetrics for depth
• Composition: add foreground passes/symmetry/rule‑of‑thirds for layered shots
• Aspect: align to the publishing channel (16:9 horizontal / 9:16 vertical)
• Seed (optional): if available, fix the seed to reproduce results; tweak one variable to compare

Industry Examples (swap nouns to reuse)

• Ads/Product showcase (16:9):

Bright tabletop scene with soft white light; a matte‑silver laptop slowly opens;
Camera pushes slightly, 35mm look, shallow depth of field;
Accentuate brushed metal and edge highlights; natural reflections; clean cool commercial grade; steady; 16:9.

• Cinematic micro‑story (16:9):

Dusk on a wooden pier, strong wind and waves; a raincoat‑wearing teen faces the wind;
Low‑angle slow push, slight handheld sway;
Backlit silhouette with rim highlights, visible volumetric light; realistic style, warm‑cool contrast;
End on the teen’s profile close‑up; 16:9.

• Illustration/animation vibe (9:16):

Anime‑style city at night with neon signs; pink‑blue palette;
Camera descends slowly from a high angle; gentle depth changes;
Cyber‑inspired illustration with clean linework; slow pacing; 9:16.

Image‑to‑Video Examples

• Portrait (9:16):

Match the subject’s hair/clothes/facial features; camera pushes from medium to chest‑up close‑up;
Slight dynamic bokeh in the background; soft backlight;
Photoreal commercial finish with cool tones; end on a gentle smile freeze; 9:16.

• Poster partial motion (16:9):

Follow the uploaded poster; keep brand elements and color palette;
Small lateral move plus push‑in; slight light sweep across the product surface to reveal material;
Clean commercial style; end on the product logo; 16:9.

Debug Checklist

• Subject drift/deformation: reinforce subject identity and key appearance; reduce style noise
• Cluttered frames: cut adjectives; ask for minimal/clean/steady; reduce scene elements
• Random camera: specify slow push/steady/small pan; avoid fast/heavy motion words
• Unwanted lighting: define cool/warm, back/side; add reflection/volumetrics/silhouette
• Weak narrative: add action + ending (e.g., look back, end on freeze)

Suggested Workflow

1. Minimal prompt to validate direction (10–20 words) → 2) Add camera/lighting/style gradually → 3) Tweak one variable per round
2. Lock the final keywords and pacing → 5) Add music, captions, and motion graphics in editing

—

Ready to create?

• Start here: Text‑to‑Video
• Animate images: Image‑to‑Video