The Complete Seedance 2 Prompt Guide

Master the art of AI video generation with our comprehensive guide to Seedance 2 prompt engineering. From basic structure to advanced multimodal techniques, learn everything you need to create stunning videos.

Table of Contents

Prompt Structure Fundamentals

The Basic Formula

Every effective Seedance 2 prompt follows a simple four-part structure:

[Subject] + [Action] + [Camera Movement] + [Style]

Example:

Subject: A sleek sports car
Action: drifting around a mountain curve at sunset
Camera: tracking shot following from the side
Style: cinematic color grading with golden hour lighting

Four Key Components Explained

1. Subject Description

The main focus of your video - person, object, animal, or scene. Be specific with details.

✓ Good"A young woman with long flowing red hair wearing a white summer dress"
✗ Vague"A person"

2. Action Description

What's happening in the scene. Use vivid verbs and sequence actions with "then".

✓ Good"gracefully walking through a field of lavender, her dress flowing in the gentle breeze"

3. Camera Movement

How the virtual camera moves and frames the subject. Essential for creating dynamic videos.

4. Style & Aesthetics

Visual treatment including lighting, color grading, artistic style, and mood. Combine multiple keywords.

Camera Movement Cheat Sheet

Camera movement is one of the most powerful tools in Seedance 2. The right camera work can transform a simple scene into a cinematic masterpiece.

Static Shots

When to use: Intimate moments, product showcases, establishing shots

Static Shot

Fixed camera, no movement. Perfect for product photography and portraits.

static shot, centered composition
Close-up

Tight framing on subject for emotional moments and detail shots.

close-up shot of the face
Wide Shot

Shows full scene/environment for establishing scenes and landscapes.

wide shot showing the entire valley
Over-the-shoulder

Camera behind subject, great for conversations and POV shots.

over-the-shoulder shot

Dynamic Movements

When to use: Action sequences, revealing scenes, following subjects

Tracking Shot

Camera follows subject through space.

tracking shot following the runner
Pan

Horizontal rotation for revealing wide scenes.

slow pan from left to right
Zoom In/Out

Lens magnification changes for focus or revealing context.

slow zoom in on the eyes
Orbit

Camera circles around subject for 360° views.

camera orbiting around the character

Writing Effective Action Descriptions

The Rule: Be Specific
✗ Vague"a person moving"
✓ Specific"a young woman gracefully walking through a field of lavender, her white dress flowing in the gentle breeze"
Multi-Action Sequences

Use "then" to separate sequential actions. Limit to 2-4 actions per prompt.

A cat sleeping peacefully on a windowsill,
then slowly waking up and stretching,
then turning to look at the camera with curious eyes

Style Keywords Library

Style keywords transform your video's aesthetic. Mix and match these to create your desired look.

Visual Styles

Realism:

photorealistichyperrealistic4K qualityhigh detailsharp focus

Cinematic:

cinematicfilm grainshallow depth of fieldanamorphic lensbokeh

Artistic:

anime stylecartoonhand-drawncel-shadedwatercolor
Lighting Conditions
golden hour lightingdramatic lightingstudio lightingmoody lightingethereal glowvolumetric light rays

Multimodal Input Techniques

Seedance 2.0 supports multimodal input, allowing you to combine text prompts with image and audio references for unprecedented control.

Image-to-Video (I2V)

Use a static image as the starting point, then describe the motion. Focus your prompt on movement, not re-describing the image.

Reference Image: Sports car photo
Prompt: "drifting around a sharp corner, tire smoke visible"
Text-to-Video (T2V)

Pure text description generates the entire video from scratch. Requires extremely detailed descriptions.

"A massive dragon with emerald scales perched atop a medieval castle tower..."

Common Mistakes & How to Fix Them

Mistake 1: Vague Descriptions
✗"A beautiful landscape"
✓"A snow-capped mountain range at sunset, with a crystal-clear alpine lake reflecting pink clouds"
Mistake 2: Overloading the Prompt

Too many elements confuse the model. Focus on 1-2 primary subjects and 2-3 secondary elements max.

Mistake 3: Ignoring Camera Movement

Without camera direction, Seedance defaults to static or random movements. Defining camera work dramatically improves quality.

Advanced Tips & Tricks

The Power of Specificity

General Rule: More specific = Better results

Generic"a car driving fast"
Specific"a red 1967 Ford Mustang convertible accelerating down Route 66, engine roaring, dust kicking up behind"

Ready to Put This Into Practice?

Now that you've mastered the fundamentals of Seedance 2 prompt engineering, it's time to create. Browse our library of battle-tested prompt templates for instant inspiration.

Browse Prompt Templates →Check Out Examples

💡 Remember: Subject + Action + Camera + Style = Perfect Prompt