✨ Write better prompts for generating clips with AI

.png)

ProdOps @ Synthesia

📌 Sora 2 is no longer available on the Synthesia platform. Any videos currently using Sora 2 assets will not be affected, but you will not be able to generate new assets with Sora 2.

The latest video generation models from OpenAI and Google are available in Synthesia so you can generate custom video clips with them. Google’s Veo 3.1 and OpenAI’s Sora 2 models translate written prompts into rich, dynamic video scenes. Every detail you include (or leave out) influences how the model generates the output.

Unlike static media, video is multi-dimensional. When you write a prompt for a video model, you’re going beyond just describing what’s happening in the video, it’s more like you’re directing a scene. That includes everything from what to show, how to move the virtual camera, how to light the scene, what sound to include, and how everything should feel.

The model doesn’t guess your intent; it only knows what you tell it. So…if your prompt is something vague (i.e. “a person in an office”) you’ll probably end up with something pretty generic. But if you define key visual elements like the subject, setting, lighting, composition, and camera motion, you’ll get a much more intentional, cohesive output.

In this post, you’ll find a breakdown of each part of a high-quality prompt. I hope you'll find this info helpful as you start to experiment with Synthesia’s Veo 3.1 and Sora 2 integrations!

1. Scene summary

Start your prompt with a clear, concise sentence that sets up the big picture before you clarify the finer details. Think of this as the headline for your video scene: what's happening, and how should it feel?

• What’s the main action?
• What mood or tone are you trying to set?
• Who is in the scene and what are they doing?

Example: “A cinematic scene capturing a filmmaker's perspective on scene direction, featuring a person thoughtfully planning a shot in a modern, well-lit studio.”

2. Camera language

Camera language tells Veo 3 how to simulate the look and feel of professional cinematography. It's not just about what's being shown—it's how the viewer experiences it.

Camera angle

This is the height or tilt of the virtual camera relative to the subject:

• Eye-level: Feels neutral, conversational.
• High-angle: Looks down on the subject, often making them feel small or vulnerable.
• Low-angle: Looks up at the subject, conveying power or drama.

Example: “A pigeon is sitting on a New York City park bench, looking downtrodden. The camera quickly moves down and tilts to a low-angle close-up shot of the pigeon's face, making him look powerful and menacing.”

Camera movement

Movement brings life to a scene. Instead of a static frame, you can describe motion to add cinematic polish:

• Pan right/left: Camera swivels horizontally.
• Whip pan: A camera panning movement fast enough to give a blurred picture.
• Tilt up/down: Camera moves vertically.
• Dolly shot: Camera physically moves closer or further away.
• Tracking shot: Follows the subject smoothly.
• Handheld: Mimics a raw, natural feel with slight motion.

Example: “Pan slowly to the right, then quickly whip pan to the right (with no sound effect) to reveal the violinist playing music.”

Framing, lens, and composition

This describes how the subject is positioned in the frame and what the "camera" sees:

• Close-up: Tight shot on a face or object, great for emotion.
• Wide shot: Shows full subject in environment.
• 50mm lens: Mimics natural human vision.
• Wide-angle lens: Captures more background with slight distortion.
• "Framed off-center with leading lines": Adds dynamic tension or interest.

Example: “Slow dolly-in on a low-angle close-up of the musician’s face, shot on a 50mm lens.”

Resources

• Learn about different framings with this great guide for camera shot terminology (+ examples) → 50+ Types of Camera Shots, Angles, and Techniques - studiobinder
• Learn about each type of camera movement you can add with this guide → Types of Camera Movements in Film Explained: Definitive Guide - studiobinder

3. Subject detail

Clearly describe who or what is the focus of the shot to ensure that characters or objects are rendered accurately/as you imagine them.

Include:

• Physical appearance (age, hair, skin tone, clothing)
• Facial expression (smiling, focused, concerned)
• Body language (sitting, pacing, relaxed posture)
• Gender and ethnicity (optional, but useful for specificity)

For character consistency across clips, copy and paste the exact same description.

Example: “A woman in her early 30s with shoulder-length curly black hair, wearing a denim jacket and a thoughtful expression.”

4. Time & setting

Describe where and when your scene takes place. This sets the environmental context and contributes to the mood.

Location

Indoor or outdoor? Urban or pastoral? Real or imagined?

• "In a bustling open-plan office"
• "In a quiet forest glade"
• "On a rooftop overlooking the city"

Time of day

This affects lighting, shadows, and energy:

• "At golden hour just before sunset"
• "Late at night with only streetlights visible"
• "Early morning with fog in the air"

Era/time period

Including this is optional, but can be useful if you want a retro or futuristic look:

• "1980s kitchen"
• "Cyberpunk city in 2075"

Example: “On a rain-slicked Tokyo street at dusk, neon signs reflecting off the wet pavement.”

5. Lighting & color grading

Lighting isn’t just technical—it’s emotional. It shapes how we feel about a scene. Color grading is the visual filter applied to the whole video to set tone and atmosphere.

Lighting types

Use film-style terminology:

• Golden hour: Soft, warm, flattering
• Rim lighting: Glowing edge behind subject
• Chiaroscuro: Deep shadows with dramatic highlights
• Overhead fluorescents: Harsh and clinical

Example: “The room is lit with a golden hour glow – soft, warm light streams in from the left, casting gentle highlights across the fur. Then, the lighting suddenly switches to overhead fluorescents – stark, cold light casts clinical shadows, flattening the mood.”

Color grading

Describes the scene’s overall color tone:

• Cool tones: Blues and grays, often used for serious or modern scenes
• Warm tones: Oranges and yellows, nostalgic or inviting
• High contrast: Deep shadows, bright highlights
• Muted colors: Low saturation, artistic or moody

Example: “Color grading is warm, dusty, and slightly desaturated to reinforce the whimsical melancholy of the scene.”

6. Motion & action

Be specific about what the subject is doing and how they move.

Movement style

• "Walks quickly with urgency"
• "Dances slowly, gracefully"
• "Turns to the camera with hesitation"

Rhythm

Describe speed and pacing:

• "In slow motion as rain falls around them"
• "Time-lapse showing the city waking up"

These motion cues add texture and realism.

Example: “She walks with purpose through puddles, coat flapping behind her in the wind.”

7. Color palette

This affects how everything in the scene appears visually—from wardrobe to lighting to background.

• "Muted earth tones"
• "High-contrast with red accents"
• "Retro pastel colors"
• "Neon cyberpunk palette"

You can also reference known film styles:

• "In the style of a Wes Anderson film with symmetrical framing and pastel colors"

Example: “The scene is heavily influenced by the visual language of Chungking Express—color grading includes rich emerald greens, saturated yellows, and glowing reds.”

8. Audio design (Sora 2 only)

Video without audio is half a story. Sora 2 lets you guide not just what’s seen, but what’s heard—and that can completely change how your scene feels.

Think of audio prompting like sound design notes in a film script: you’re describing the mood, texture, and rhythm of the soundscape.

Sound categories to define:

1. Ambient sound What’s the world around the viewer doing?

• “Soft hum of airplane engines and quiet chatter in the cabin.”
• “Wind whistling through trees with distant city noise.”

2. Foley / action sounds Describe sounds tied to motion or physical action.

• “Footsteps crunching on gravel.”
• “Metal buckle clicks shut.”

3. Emotional tone or mix Set how immersive or dominant sound should feel:

• “Subtle, background sound design to support calm mood.”
• “Loud, echoing soundscape to heighten tension.”

4. Music or score (optional) If you want music, say what kind and what role it plays.

• “Minimal ambient synth underscores the scene.”
• “Playful jazz cues add a lighthearted tone.” If you want no music: say that too—“no music, only natural sound.”

5. Silence and contrast Moments of quiet are powerful. You can note this:

• “Sound drops out completely as the camera lingers on the subject.”

Example:“Interior of a commercial plane during boarding. Soft ambient noise of air vents and seatbelt clicks. Murmured conversation fades under a calm, neutral hum. No music—only realistic aircraft sounds.”

Additional tips

• Instead of saying what you don't want, be specific about what you do want. For example, "an empty room with white walls" is better than "no furniture or clutter."
• Start your prompt with a strong subject reference, like "Close-up of a man in a red jacket" or "Medium shot of a teacher at the front of a classroom."
• Add lighting details like "soft key light from the left," or describe time of day to influence mood and prevent the scene from looking flat or lifeless. Avoid vague terms like "bright" or "dark."

Recap

To get the best results from these video models, treat your prompt like a scene from a movie script. The more intentional your details, the more cinematic and cohesive your video will be. Here’s what to remember:

• Start with a clear scene summary. What’s happening, and what’s the mood?
• Use camera language. Describe angles, movement, and framing like a director.
• Clearly define your subject. Be specific with appearance, emotion, and posture.
• Set the scene. Describe the location, time of day, and overall vibe.
• Guide lighting and color grading. Use cinematic terms to set tone and atmosphere.
• Add motion with intention. Describe how characters or the camera should move.
• Mention a color palette or visual reference . Create visual cohesion and style.

Ready to experiment? Start with a single visual idea—like a specific lighting style or emotional tone—and build around it. Oh, and don’t forget to share your favourite prompts in the thread!