Direct a scene with visual references using Seedance 2.0. ByteDance’s multimodal video model combines creative context with audiovisual generation; explain what each supplied reference contributes.
Explore the tasks shown by these actual Seedance 2.0 results. Copy a prompt to try in the generator, or open the linked case for its saved setup. Supply your own images or clips when the request uses references.
Graphic Product Variations
Develop a graphic product sequence from its supplied visual context, with arrangement and visual treatment guiding the request.
0–2s: Rapid four-panel flash cuts. Red, pink, purple, and leopard-print bows freeze in sequence, with close-ups of the satin sheen and the “chéri” logo. Voiceover: “Create endless beauty with Chéri magnetic ribbons!”
3–6s: Close-up of the silver magnetic clasp snapping shut with a “click,” then gently pulled apart, highlighting the silky texture and convenience. Voiceover: “Fasten it in just 1 second and complete the ultimate style!”
7–12s: Quick cuts of wearing scenarios: the burgundy one pinned to a coat lapel for a full-on commuter vibe; the pink one tied on a ponytail for a sweet-girl outing; the purple one on a bag strap for a niche, elevated look; the leopard-print one on a suit lapel for a bold, fierce aura. Voiceover: “From coats and bags to hair accessories—create versatile, personality-filled styles!”
13–15s: All four bows displayed side by side. Brand line: “chéri, bringing you instant beauty!”
Build a character scene around a reflective composition, using the original inputs and prompt for the visual direction.
Based on <User Portrait>, an exquisitely refined woman positioned at a private bathroom vanity, bathed in intense afternoon sunlight. Her posture is relaxed and languid, yet retains an unmistakable sense of elegance
Develop a fantasy action sequence around the characters and battle environment described in the request.
A colossal armored baboon charges through a burning Maasai village, it picks up and throws a maasai warrior into a burning hut, Maasai warriors are running away in fear amid a storm of fire and debris.
Seedance 2.0 is ByteDance’s unified multimodal audio-video model. It uses text, image, audio, and video context for directed generation and reference-based creation.
Multimodal Creative Context
The 2.0 release supports creative direction from text, image, video, and audio inputs. Reference roles can distinguish appearance, movement, or other scene information.
Picture and Sound Together
Seedance 2.0 is an audiovisual model. Dialogue, ambience, and visible performance can be planned as parts of one scene.
Reference-Led Direction
References can guide more than a single opening composition. The current form determines which media and modes are accepted for a request.
WORKFLOWS
Choose Your Starting Input
Use the Seedance 2.0 input mode that matches the material you have.
Direct a Scene with References
Describe the scene and add reference media through the available controls. Identify what each input supplies, such as the subject, setting, motion, or sound.
PROMPT GUIDE
How to Direct Seedance 2.0 with a Prompt
Learn from the actual instructions behind these results. Each example keeps its complete original prompt and available inputs. For a new request, replace its subject and reference assignments with your own.
Put the Events in Order
Describe what happens first and how the next action or reveal follows it. In this case: “The character then starts shouting loudly, in an emotional and slightly comedic way, with facial expressions and gestures referencing Image 3.”
Use Image 1 as the opening frame. The camera slowly rotates and pushes in. The character suddenly looks up. The facial features should be based on Image 2. The character then starts shouting loudly, in an emotional and slightly comedic way, with facial expressions and gestures referencing Image 3. After that, the character transforms into a bear, with the body form based on Image 4.
Establish the Visual Starting Point
The request defines the subject and setting. Use it to study scene description rather than treating it as evidence of a particular motion or sound control. In this case: “A nature documentary about an otter flying an airplane”
A nature documentary about an otter flying an airplane
Map the Camera Route
Name the viewpoint, the path through the scene, and where the shot should end. In this case: “First-person perspective from a playful Labrador secretly hiding its favorite tennis ball under the couch after a game of fetch.”
First-person perspective from a playful Labrador secretly hiding its favorite tennis ball under the couch after a game of fetch. The dog quietly watches the confused owner search the house while pretending not to know where it is. Hours later, when the owner finally sits down to relax, the dog proudly retrieves the hidden ball and drops it directly onto their lap, ready to play again. Authentic dog behavior, playful pacing, natural indoor lighting, realistic GoPro motion.
Specify the Intended Speech or Song
Identify the speaker or performance and its relationship to the visible scene. Check the delivered audio separately from the picture. In this case: “The image in Figure 1 is rewritten in a comic strip format from left to right and top to bottom, maintaining consistency between the characters' dialogue and the image.”
The image in Figure 1 is rewritten in a comic strip format from left to right and top to bottom, maintaining consistency between the characters' dialogue and the image. Special sound effects are added during scene transitions and key plot points, resulting in an overall humorous style.
HOW TO USE
Create Your First Seedance 2.0 Video
Start in the generator on this page. To reuse an example, click Copy Prompt, return to the description field, and adapt the prompt to your own inputs.
Provide the Scene and Inputs
Write your brief and add the reference media required by the selected mode. Explain the role of each supplied input.
Choose the Output Options
Choose an offered duration, aspect ratio, and other available output options. Check the form’s displayed points before submitting.
Generate and Review
Click Generate and review the resulting video. Check movement, framing, continuity, and any delivered audio. Adjust your prompt or references for the next attempt.
Compare the documented model roles and the inputs and options currently listed in WeShop. Model names or matching settings do not by themselves establish equal output quality or a fixed generation-time difference.
Dimension
Seedance 2.0
Seedance 2.0 Mini
Seedance 2.5
Model role
Multimodal creative direction and audiovisual generation
Mini selection for multimodal creation
Longer stories, reference interpretation, and directed editing
Inputs listed for the page
Direct a Scene with References
Direct a Scene with References
Direct a Scene with References
Current WeShop options
Duration: 4–15 seconds
Duration: 4–15 seconds
Duration: 4–30 seconds
Model role
Seedance 2.0
Multimodal creative direction and audiovisual generation
Seedance 2.0 Mini
Mini selection for multimodal creation
Seedance 2.5
Longer stories, reference interpretation, and directed editing
Inputs listed for the page
Seedance 2.0
Direct a Scene with References
Seedance 2.0 Mini
Direct a Scene with References
Seedance 2.5
Direct a Scene with References
Current WeShop options
Seedance 2.0
Duration: 4–15 seconds
Seedance 2.0 Mini
Duration: 4–15 seconds
Seedance 2.5
Duration: 4–30 seconds
When to Consider Seedance 2.0
Match the Model to Your Task
Consider 2.0 for a multimodal creative brief. Compare Mini and 2.5 using their actual input controls and durations, rather than applying the newer version’s limits to 2.0.
MORE VIDEO MODELS
Explore More Video Models
Open a related model page to review its capabilities, examples, and current generation options.
Answers about this version, its inputs, and how to get started.
What is Seedance 2.0?
Seedance 2.0 is ByteDance’s unified multimodal audio-video model. It uses text, image, audio, and video context for directed generation and reference-based creation.
What is a good task to consider it for?
Consider 2.0 for a multimodal creative brief. Compare Mini and 2.5 using their actual input controls and durations, rather than applying the newer version’s limits to 2.0.
What do I need to provide?
Direct a Scene with References. Describe the scene and add reference media through the available controls. Identify what each input supplies, such as the subject, setting, motion, or sound.
Which output settings are offered here?
Duration: 4–15 seconds
How do I try an example prompt?
Click Copy Prompt, return to the generator, and paste it into the description field. Add your own required references and update their roles. Linked case previews and Use This Prompt open the saved case; they do not automatically fill this page’s form.
What should I know about this version?
Do not inherit Seedance 2.5’s 30-second single-generation limit or standalone audio-reference behavior from its newer release.
