How to Use Seedance 2.0: Prompts, References & Examples

Not sure whether to use Seedance 2.0 text-to-video or image-to-video? Many first-time users also struggle with writing the right Seedance 2.0 prompt, using @Video references correctly, and keeping characters or products consistent. Choosing the right workflow before generating can prevent wasted attempts and give you more control over the final video. You can try Seedance 2.0 for free and follow along with this guide.

For example, you can use one image to define a character, another to preserve a product, a video to demonstrate motion or timing, and an audio file to guide the final reveal.

In this guide:

• What Is Seedance 2.0?
• Seedance 2.0 Capabilities at a Glance
• How to Use Seedance 2.0 Step by Step
• Which Seedance 2.0 Workflow Should You Use?
• How to Improve Seedance 2.0 Results
• Frequently Asked Questions
• Create Your First Seedance 2.0 Video

What Is Seedance 2.0?

Seedance 2.0 is ByteDance's multimodal AI video generation model. It supports text, image, video, and audio inputs. Its main advantage is not simply the number of inputs it accepts. The real value comes from assigning a clear creative role to each input.

For example, you can use one image to define a character, another image to preserve a product, a video to demonstrate camera movement, and an audio file to guide the timing of a reveal. Your prompt then explains how these references should work together.

Seedance 2.0 Capabilities at a Glance

According to ByteDance Seed's official Seedance 2.0 launch information, the model uses a unified multimodal audio-video generation architecture and supports four input types: text, image, video, and audio.

A single task can include up to nine images, three video clips, and three audio clips alongside written instructions. The model also supports up to 15-second multi-shot audio-video output, synchronized dual-channel sound, video extension, and editing.

Available duration, resolution, reference limits, and editing options may vary by platform, account, and generation mode.

How to Use Seedance 2.0 Step by Step

Using Seedance 2.0 involves three main steps: describing the video, choosing the output settings, and reviewing the generated result.

1. Enter a Prompt and Add References

Begin with a clear description of what should happen in the video. Include the subject, primary action, setting, shot type, visual style, and sound when they are relevant.

Seedance 2.0 Prompt Example:

A silver sports car drives along a coastal road at sunset. The camera follows from a low rear angle before moving alongside the car. Warm cinematic lighting, realistic wheel movement, reflections across the bodywork, ocean wind, and subtle engine sound.

You can then upload optional image, video, or audio references. Use @ mentions to assign each file a clear role:

• Use @Image to define a character, product, location, composition, or style.
• Use @Video to guide movement, camera direction, timing, choreography, or transitions.
• Use @Audio to guide dialogue, music, rhythm, sound effects, or ambience.

A multimodal prompt might look like this:

Keep the character's face and clothing from @Image1. Follow the slow circular camera movement from @Video1 and synchronize the final product reveal with the main beat in @Audio1.

Avoid uploading a reference without explaining what the model should take from it. A Seedance 2.0 @Video reference becomes more useful when you specify whether it controls motion, shot timing, transitions, or another visible behavior.

2. Choose Your Video Settings

Select the generation settings available on the platform, which may include:

• Duration
• Aspect ratio
• Resolution
• Audio generation

Choose the aspect ratio according to where you plan to publish the video:

• Use 16:9 for cinematic scenes, websites, presentations, and YouTube.
• Use 9:16 for TikTok, Instagram Reels, YouTube Shorts, and mobile advertising.
• Use 1:1 for product posts and flexible social media publishing.

Resolution affects output clarity, but it does not correct an unclear prompt or conflicting references. Test the composition and primary action before using additional resources on a higher-resolution version.

If audio is enabled, describe the sound you want rather than relying on the model to infer everything from the visuals. Mention dialogue, ambience, music, or important sound effects directly in the prompt.

3. Generate, Preview, and Download

Click Generate and follow the task through the queue and generation stages.

When the result is ready, watch the entire video instead of judging it from the first frame. Check whether:

• The requested action takes place
• The character or product remains recognizable
• The motion looks physically believable
• The camera follows the requested direction
• The sound matches the visuals
• Each shot contributes to the intended story

If one part is incorrect, identify the failed instruction before rewriting the whole prompt. For example, if the character looks correct but the camera moves too quickly, keep the identity reference and change only the camera instruction.

This controlled approach makes it easier to understand which prompt or reference changes produce better results.

Which Seedance 2.0 Workflow Should You Use?

The best input method depends on what you already have and how much control the finished video requires.

Use Text-to-Video to Explore a New Idea

Use Seedance 2.0 text-to-video when you have a concept but do not need to preserve a specific character, product, or composition.

This workflow works well for:

• Cinematic concepts
• Fantasy environments
• Mood and lighting experiments
• Short narrative ideas
• Early advertising concepts
• Social media visuals

Text-to-video gives Seedance 2.0 more freedom to interpret the scene. This makes it useful for exploration, but the appearance of characters and locations may vary more than in a reference-based workflow.

Seedance 2.0 Prompt Example:

A lone astronaut walks through a field of silver grass beneath two moons. The camera tracks beside the astronaut in a steady wide shot as the wind moves through the landscape. Dreamlike cinematic lighting, realistic walking motion, soft ambient music, and distant wind.

This prompt works because it focuses on one subject, one action, and one consistent visual atmosphere.

Best for: Starting from an idea and discovering a visual direction.

Use Image-to-Video When Visual Identity Matters

Use Seedance 2.0 image-to-video when the appearance of a character, product, artwork, or location must remain recognizable.

It is especially useful for:

• Product advertising
• Character animation
• Portrait animation
• Fashion content
• Artwork and illustration animation
• Branded social media content
• Architectural visualization

The image establishes what the scene should look like. The Seedance 2.0 image-to-video prompt should then focus on what changes over time: subject movement, camera behavior, environmental motion, lighting changes, and sound.

Seedance 2.0 Prompt Example:

Use @Image1 as the opening composition. Keep the bottle shape, label placement, materials, proportions, and colors unchanged. Water droplets move slowly across the surface while the camera makes a gentle clockwise orbit. Premium studio lighting with subtle water and glass sound effects.

A common mistake is repeating details already visible in the image without describing the animation. Focus on movement and timing instead of rewriting the image as text.

Best for: Preserving a recognizable subject while adding controlled motion.

Use Multimodal References for Precise Creative Control

Use multimodal references when different parts of the video need to follow different source materials.

For example:

• An image defines the character.
• A second image preserves the product.
• A video demonstrates the desired movement.
• An audio file controls rhythm or atmosphere.
• The prompt explains how the references interact.

Seedance 2.0 Prompt Example:

Keep the character's face, hairstyle, and clothing from @Image1. Preserve the product design from @Image2. Follow the hand movement and camera timing from @Video1. Synchronize the final product reveal with the main beat in @Audio1.

This workflow offers the most control, but it also requires the clearest instructions. Each reference should solve a specific creative problem.

Using more references does not automatically create a better video. If the images, videos, and audio communicate different styles or pacing, the model may need to decide which direction to prioritize.

Best for: Advertising, character-driven stories, product videos, choreography, and scenes that require coordinated visual and audio control.

How to Improve Seedance 2.0 Results

Better Seedance 2.0 results usually come from clearer creative decisions rather than longer prompts. The goal is to reduce ambiguity while giving the model enough information to construct the scene.

Write a Clear Seedance 2.0 Prompt

A weak prompt often consists of unrelated style words:

Cinematic, realistic, dramatic, beautiful lighting, smooth movement, high quality.

These words influence appearance but do not explain what happens.

A stronger prompt connects the subject, action, setting, camera, style, and sound:

A woman in a red coat walks through a rain-soaked neon street at night. The camera tracks backward in a steady medium shot as reflections move across the pavement. Cinematic cyberpunk lighting, natural walking motion, light rain ambience, and distant city traffic.

Use this prompt formula:

Subject + Action + Setting + Shot Design + Visual Style + Sound

Better Seedance 2.0 Prompt Example:

[Subject] performs [specific action] in [setting]. Show the scene using [shot type, framing, or movement] while [secondary action or environmental change] occurs. Keep [important identity or product details] consistent. Use [lighting and visual style] with [sound, dialogue, music, or ambience].

For reference-based generation, add:

Use @Image1 for [character, product, location, or style]. Follow [movement, timing, camera path, or interaction] from @Video1. Use @Audio1 for [rhythm, dialogue, sound effect, or atmosphere].

Place essential instructions near the beginning of the prompt. A useful priority order: what must happen → what must remain consistent → how the scene should be filmed → how it should look and sound.

Start with One Primary Action

A short AI video cannot communicate unlimited events clearly. Begin with one main action and one supporting visual change.

Instead of:

The character runs, turns, opens a door, enters a room, speaks to another person, and looks through a window.

Try:

The character approaches the door and slowly opens it as the camera moves closer from behind.

Once the primary action works, add secondary movement or another shot.

Use References with Clear Roles

The maximum number of supported references is a technical limit, not a recommended starting point. Start with the minimum references needed to control the most important part of the video, then add another only when it solves a specific creative problem.

Do not expect Seedance 2.0 to determine automatically which part of a reference matters most. Instead of "Make the video like @Video1," write:

Follow the slow circular camera path from @Video1, but replace the original subject with the product from @Image1.

A useful division of responsibilities:

• Images control appearance and composition.
• Videos control motion and camera behavior.
• Audio controls timing and atmosphere.
• The prompt connects everything into one scene.

Change One Variable at a Time

When refining a result, avoid changing the action, references, framing, style, and settings at the same time. Use a controlled workflow:

1. Confirm the subject and composition.
2. Test the primary action.
3. Refine the shot and motion.
4. Add sound and secondary details.
5. Increase output quality if needed.

Tip: Start with one subject, one primary action, and the minimum number of references required. Confirm the basic scene works before adding more controls.

Frequently Asked Questions

Where can I use Seedance 2.0?

Seedance 2.0 is available through several platforms. You can start with Seedance AI Video Generator, a free Seedance 2.0 platform with no sign-up required and daily free generations. It is also accessible on ByteDance's official Doubao and Jimeng apps, through the Seedance API for developers, and on other third-party platforms that have integrated the model. Check your chosen platform for current access options and availability in your region.

Is Seedance 2.0 free to use?

Seedance 2.0 is a model rather than a standalone product with a fixed price. Whether you can use Seedance 2.0 free depends on the platform you access it through. Many platforms offer free trial credits or a limited number of free generations when you sign up. For example, Seedance AI Video Generator provides free daily access with no registration required. After free quotas are used, pricing typically follows a per-generation or subscription model. Visit your chosen platform's pricing page for the most up-to-date information.

How long can Seedance 2.0 videos be?

According to ByteDance's official specifications, Seedance 2.0 supports video output up to 15 seconds with multi-shot generation. The exact duration available to you may vary depending on the platform, your account level, and the generation mode you select. Some platforms also support video extension to lengthen a completed clip.

What is the difference between Seedance 2.0 text-to-video and image-to-video?

Text-to-video starts from a written prompt and gives the model more freedom to interpret the scene. It works well for exploring concepts when you do not need to preserve a specific look. Image-to-video starts from a reference image and keeps the subject, product, or composition recognizable while adding motion. Choose text-to-video for creative exploration and image-to-video when visual identity matters.

Can I use Seedance 2.0 videos for commercial projects?

Commercial usage rights depend on the platform you use to access Seedance 2.0 and your subscription level. Most platforms allow commercial use under paid plans, but restrictions may apply to free-tier outputs. Review the terms of service on your chosen platform before using generated videos in advertising, products, or client work.

Create Your First Seedance 2.0 Video

Now you know how to use Seedance 2.0: choose text-to-video to explore a new idea, image-to-video to preserve a character or product, or multimodal references for full control over motion, camera, and sound. Start with one clear action and only the references you need, then refine one variable at a time.

Need more examples? Browse the Seedance AI inspiration gallery to see real Seedance 2.0 video creations, preview results, and copy the exact prompts other creators used.

Ready to try Seedance 2.0 free? Start creating on Seedance AI Video Generator — no sign-up required, free daily generations available. Use the prompt formula above, pick a workflow, and turn your idea into a cinematic AI video in minutes.