The Sora 2 Prompt Generator is a powerful tool designed to help you craft structured, cinematic prompts tailored for OpenAI’s Sora 2 video models.

It simplifies the process of creating production-ready prompts by gathering detailed inputs—such as subjects, cinematography, lighting, and dialogue—and outputs clean, structured prompts alongside API parameters.

This ensures seamless integration into your Sora workflow.

Key Takeaways

Structured Inputs

• Capture essential elements like:
  • Model
  • Size
  • Duration
  • Shot craft
  • Lighting
  • Color
  • Continuity
• Inputs are organized into discrete fields for clarity.

Separation of Concerns

• Keeps the prose prompt and API parameters separate to avoid cross-contamination.
• Ensures clean, focused outputs.

Dynamic Builders

• Add or remove elements like:
  • Subjects
  • Action beats
  • Dialogue lines
• Adjust inputs on the fly for flexibility.

Model-Aware Sizes

Automatically updates size options based on the selected model:

• sora-2-pro unlocks additional resolutions.

Copy-Ready Outputs

• One-click copy functionality for:
  • Composed prose prompt
  • API JSON parameters

Key Features of Sora 2 Prompt Generator

Model, Size, and Duration Controls

• Model Selection: Choose between sora-2 and sora-2-pro.
• Size Options:
  • sora-2: 1280x720, 720x1280
  • sora-2-pro: Adds 1024x1792 and 1792x1024
• Duration: Select clip lengths of 4, 8, or 12 seconds.

Example API JSON:

{
  "model": "sora-2-pro",
  "size": "1280x720",
  "seconds": 8,
  "input_reference": "https://your-cdn.com/ref-frame-1280x720.png"
}

Structured Scene and Style Inputs

• Define the creative direction with fields for:
  • Style: e.g., “neo-noir realism”
  • Mood: e.g., “tense but intimate”
  • Genre/Era: e.g., “1950s crime”
  • Theme: e.g., “trust and betrayal”

Location, Time, and Atmosphere

• Specify the setting and ambiance:
  • Location: Where the scene takes place (e.g., “rooftop overlooking neon city”).
  • Time of Day: Golden hour, midnight, etc.
  • Weather/Atmosphere: Fog, drizzle, smog, snow.
  • Ambience: A single meta cue for soundscapes (e.g., “distant city hum”).

Dynamic Subjects Builder

• Add or remove multiple subjects with details like:
  • Name/Role: e.g., “Courier”
  • Description: e.g., “lean, nervous energy; damp jacket; hair matted by drizzle”
  • Wardrobe: e.g., “dark trench coat, leather courier bag”
  • Props: e.g., “sealed envelope, flip phone”

Example subject block:

{
  "name_or_role": "Courier",
  "description": "lean, nervous energy; damp jacket; hair matted by drizzle",
  "wardrobe": "dark trench coat, leather courier bag",
  "props": ["sealed envelope", "flip phone"]
}

Action Beats (Time-Aware)

• Add concise, time-aware action lines:
  • One subject action per shot.
  • Keep beats realistic for 4/8/12-second durations.
• Example:
  • “He glances over the edge; neon flickers reflect on his face.”
  • “A gust flaps the coat; envelope edges show.”

Dialogue Support

• Add dialogue lines with labeled speakers:
  • Example:
    • “Courier: ‘You’re late. That’s not like you.’”
    • “Voice on phone: ‘Traffic. Don’t open it.’”

Cinematography Controls

• Define cinematic elements:
  • Camera Shot: e.g., “Wide establishing,” “Medium two-shot,” “Closeup.”
  • Depth of Field: e.g., “shallow DOF, creamy bokeh.”
  • Lens/Filtration: e.g., “35mm anamorphic, 1/8 Black Pro-Mist.”
  • Camera Motion: e.g., “slow dolly in; subtle handheld sway.”

Example cinematography block:

Cinematography:
  Camera shot: Medium closeup
  Mood: tense, intimate
  Depth of field: shallow, creamy bokeh
  Lens and filtration: 35mm anamorphic, 1/8 Black Pro-Mist
  Camera motion: slow dolly-in

Lighting and Color Discipline

• Lighting Description: Define sources, direction, and quality.
• Palette Anchors: Stabilize color with 3–5 concrete anchors.
• Example:
  • Lighting: “soft window key camera-left; sodium street spill; cool rim from neon signage.”
  • Palette anchors: “neon magenta, sodium orange, slate blue, carbon black.”

Continuity Notes and Negative Cues

• Continuity: Maintain consistency across shots.
  • Example: “Keep courier’s trench coat damp; consistent bag and envelope.”
• Negative Cues: Prevent unwanted artifacts.
  • Example: “no cartoonish lighting, no fisheye distortion.”

Input Reference (Optional)

• Provide a URL to visually anchor frame 1.
• Example: https://your-cdn.com/ref-frame-1280x720.png

Remix Change (Optional)

• Request a single change iteration to the previous prompt.
• Example: “Switch to rainy night; keep camera motion and subjects the same.”

Clean Outputs and Copy Buttons

• Outputs include:
  • Prose Prompt: Preformatted for readability.
  • API JSON: Ready for Sora API integration.
• Copy buttons for both outputs ensure quick usage.

How to Use Sora 2 Prompt Generator

1. Choose Model: Select sora-2 or sora-2-pro.
2. Pick Size: Size options auto-update based on the model.
3. Set Duration: Choose 4, 8, or 12 seconds.
4. Style & Tone: Fill in Style, Mood, Genre/Era, and Theme.
5. Place the Scene: Define Location, Time of Day, and Weather/Atmosphere.
6. Add Subjects: Add Name/Role, Description, Wardrobe, and Props.
7. Cinematography: Specify camera shot, depth of field, lens/filtration, and motion.
8. Action Beats: Add short, time-aware beats.
9. Dialogue (Optional): Add concise lines with speakers.
10. Lighting & Palette: Describe lighting and set 3–5 palette anchors.
11. Ambience (Optional): Add minimal audio cues.
12. Continuity & Negative Cues (Optional): Lock identities/props; prevent artifacts.
13. Input Reference (Optional): Paste a URL to anchor frame 1.
14. Remix Change (Optional): Request one succinct change.
15. Generate: Click Generate to compose the prompt and API JSON.
16. Copy Outputs: Use the copy buttons to paste into your Sora workflow.

Best Practices

• Write Time-Aware Beats: Keep actions plausible within 4/8/12 seconds.
  • Good: “He glances over the edge; neon flickers reflect.”
  • Not great: “He debates his life choices.”
• One Camera Move, One Subject Action per Shot: Avoid overloading beats.
  • Good: “Slow dolly-in as courier tucks envelope inside coat.”
  • Not great: “Dolly + whip pan + crane + chase.”
• Use Concrete Visual Language: Models respond better to specific nouns/verbs.
  • Good: “wet asphalt; sodium glow; steam vents.”
  • Not great: “cool vibes; cinematic.”
• Use Palette Anchors (3–5): Stabilize color across shots.
  • Example: “neon magenta, sodium orange, slate blue.”
• Control Lighting Explicitly: Avoid accidental flatness.
  • Example: “soft window key camera-left; neon rim.”
• Maintain Continuity: Prevent identity drift and prop changes.
  • Example: “Courier’s trench stays damp; same leather bag.”
• Use Negative Cues Sparingly: Avoid unintended artifacts.
  • Example: “no fisheye, no cartoon lighting.”
• Input References Anchor Frame 1: Match the image size to your resolution.
• Remix One Change at a Time: Keep iterations controllable.
  • Example: “Change to rainy night only.”

Frequently Asked Questions (FAQs)

What’s the difference between sora-2 and sora-2-pro?

• sora-2-pro supports more resolutions and broader capabilities.

Which resolutions can I choose?

• sora-2: 1280x720, 720x1280
• sora-2-pro: Adds 1024x1792, 1792x1024

How long can clips be?

• 4, 8, or 12 seconds.

Does this tool render the video?

• No, it composes prompts and API parameters for Sora API rendering.

What’s input_reference?

• An optional image URL to anchor frame 1.

How do I get variations?

• Use “Remix change” for controlled iterations.

Is there a rate limit?

• Yes, requests are rate-limited per IP.