Combine images, videos, audio, and text to produce cinematic videos with precise references, seamless extension, and natural language control.
Multi-Modal Input · Reference Anything · 4-15 Seconds · Watermark-Free
Core model
Seedance 3.0
Input types
Text · Image · Video · Audio
Best for
Marketing · Education · Storytelling
Build multimodal scenes in one workspace. Bring images, video, and audio as references, then describe the subject, action, camera, and sound in the prompt and generate.
Image 1
@Image 1, friendly and joyful atmosphere, then the screen gradually blurs, displaying the text "Joy is in Seedance" in the center.
Image 1
@Image 1 are jogging on a school track in sportswear. The girl looks at the boy and says confidently: "We can definitely do it!" Cut to a close-up of the boy, who hesitantly replies: "Are you sure?" Cut back to a medium close-up of the girl, who says cheerfully: "Yes!" The mood is bright and determined. Speech bubbles appear around the speaking character with the dialogue.
Image 1,2,3
Storyboard
Seedance 3.0 is the multimodal generation model in the Seedance line. The idea it is built around is the one you can practise right here: give the model the images, video, and audio you want it to learn from, then use a text prompt for the subject, action, camera, mood, and sound — instead of forcing one long prompt to describe every visual detail.
Use text to describe the result, images to define a person, product, or visual style, video to demonstrate movement and camera rhythm, and audio to guide speech, music, or atmosphere. Giving each input a clear role reduces guesswork and makes the first generation easier to direct.
Separate references can establish a character’s appearance, a product’s physical details, the environment, lighting, movement, and sound. This gives the model clearer visual boundaries and helps related shots feel like parts of the same project instead of unrelated clips.
When an idea needs more room, a single generation carries an opening, development, and ending across up to 30 seconds. Native 4K output covers the work where detail has to hold — products, lighting, typography, and close-ups — so you finish with a fuller video to keep editing rather than a clip to rebuild.
Seedance 2.0 already supports multimodal creation. Seedance 3.0 expands that foundation on four fronts: longer scenes, a much larger reference stack, stronger visual detail, and scene planning that no longer has to fit inside a short test clip.
| Longer Single Generations: Up to 30 Seconds | Seedance 2.0 is configured for 4–15-second generation. Seedance 3.0 extends a single generation to 30 seconds, leaving room for a setup, camera movement, primary action, transition, and ending beat. A longer canvas makes it possible to plan one coherent sequence instead of splitting the same idea across several short tests. |
|---|---|
| A Larger Multimodal Reference Stack | Seedance 2.0 supports up to 9 images, 3 videos, and 3 audio files. Seedance 3.0 raises this to 50 materials: up to 30 images, 10 videos, and 10 audio files. A stack that size lets character, product, environment, movement, and sound direction each occupy their own clearly defined source group. |
| Native 4K with Stronger Visual Detail | Native 4K output is there for the work that has to survive review: product texture, close-ups, launch media, lighting, and cinematic framing all hold up to inspection at a resolution where detail does not quietly disappear. |
| More Complete Scene Planning | The longer duration and larger reference stack make it easier to plan a complete scene instead of a short test clip. Seedance 3.0 also supports 3D spatial pre-visualization and camera blocking when subject placement, camera paths, and movement direction matter to the concept. |
Browse scenes built in the workspace, each shown with the prompt and reference material behind it.
The capabilities behind the workspace, and how each one changes the way a shot gets built.
One generation runs up to 30 seconds, with room for a setup, a main action, a transition, and a closing beat without stitching separate clips together.
Bring up to 50 materials into one scene — 30 images, 10 videos, and 10 audio files — enough to give character, product, environment, motion, and sound each their own source group.
Render in native 4K for work where detail has to survive review: product texture, the printed text on packaging, fabric weave, and close-ups that dissolve at lower resolution.
Point at what you want carried over — a camera path, a performance, a lighting mood — instead of packing every visual detail into one long prompt.
Give a character or product its own reference group and repeat its defining details in the prompt, so related shots read as one project rather than a set of unrelated clips.
Built for camera blocking and spatial previs, where subject placement and camera paths need to be argued out before anyone books a shoot day.
Labels like "social media" or "e-commerce" hide the differences that matter in production. These are the specific jobs a 30-second canvas and a deep reference stack are built to serve.
Earn the first two seconds: a transformation, a surreal effect, a music-led cut. Images fix the subject, a motion reference carries the effect, audio sets the rhythm.
Build a product-led video from authorized packshots and material details. Show what the item is and how it is used, without effects that outshine the product.
Hold one character across scenes by giving appearance, wardrobe, and movement their own references, then explore styles without redrawing the character each time.
Test composition, camera paths, transitions, and the closing frame before a shoot is booked, so the discussion happens over footage rather than a deck.
Let an audio reference set the timing while image and video sources establish the performer, the location, and the camera language around them.
Turn a procedure into a watchable sequence where the steps stay in order and the subject stays recognizable from the first frame to the last.
Move stills into motion so a room reads as a space: where the light falls, how the layout connects, what the finish actually looks like.
Take panels that already exist and put them in sequence and in motion, to find out which beats hold and which need reworking before production.
Keep wardrobe, product, environment, and lighting direction in separate reference groups so a brand's visual rules survive from one film to the next.
A marketplace product demo, a vertical hook, and a campaign previs need different aspect ratios, pacing, and detail levels. Settle this before writing a word of prompt.
Let one source define the character, another the product, another the camera rhythm. Separated sources leave far less to guesswork than a single crowded image.
Subject, setting, action, camera, light. State when the camera moves and what the last frame should say, then change one variable at a time as you refine.
Six working creators on what they get out of the open workspace now, and what the 30-second canvas changes for them.
“Thirty seconds in one pass is the number that matters to me. At fifteen I am always cutting the ending off; at thirty the reveal and the closing frame finally fit in the same take.”
Marcus Rodriguez
Filmmaker
“Right now I burn most of a reference slot just describing the character. Splitting appearance, wardrobe and motion into separate sources is what I actually want from a deeper stack.”
Jessica Liu
Animation Director
“We stopped writing one enormous prompt and started pointing at things instead — this shot's lighting, that clip's camera move. The team argues less and ships more.”
Emily Watson
Creative Director
“Thirty seconds in one pass is the number that matters to me. At fifteen I am always cutting the ending off; at thirty the reveal and the closing frame finally fit in the same take.”
Marcus Rodriguez
Filmmaker
“Right now I burn most of a reference slot just describing the character. Splitting appearance, wardrobe and motion into separate sources is what I actually want from a deeper stack.”
Jessica Liu
Animation Director
“We stopped writing one enormous prompt and started pointing at things instead — this shot's lighting, that clip's camera move. The team argues less and ships more.”
Emily Watson
Creative Director
“Thirty seconds in one pass is the number that matters to me. At fifteen I am always cutting the ending off; at thirty the reveal and the closing frame finally fit in the same take.”
Marcus Rodriguez
Filmmaker
“Right now I burn most of a reference slot just describing the character. Splitting appearance, wardrobe and motion into separate sources is what I actually want from a deeper stack.”
Jessica Liu
Animation Director
“We stopped writing one enormous prompt and started pointing at things instead — this shot's lighting, that clip's camera move. The team argues less and ships more.”
Emily Watson
Creative Director
“Thirty seconds in one pass is the number that matters to me. At fifteen I am always cutting the ending off; at thirty the reveal and the closing frame finally fit in the same take.”
Marcus Rodriguez
Filmmaker
“Right now I burn most of a reference slot just describing the character. Splitting appearance, wardrobe and motion into separate sources is what I actually want from a deeper stack.”
Jessica Liu
Animation Director
“We stopped writing one enormous prompt and started pointing at things instead — this shot's lighting, that clip's camera move. The team argues less and ships more.”
Emily Watson
Creative Director
“Changing one variable at a time is the whole discipline. Once I stopped rewriting the entire prompt between attempts I could actually tell what was improving.”
Mohammed Hassan
Digital Artist
“Letting the audio set the timing while images handle the performer and the location is the closest thing to how I plan a real shoot.”
Alex Turner
Music Video Director
“More usable seconds per generation means more material to judge pacing against. Native 4K matters for the same reason — I need detail that survives the review, not just the preview.”
Olivia Martinez
Video Editor
“Changing one variable at a time is the whole discipline. Once I stopped rewriting the entire prompt between attempts I could actually tell what was improving.”
Mohammed Hassan
Digital Artist
“Letting the audio set the timing while images handle the performer and the location is the closest thing to how I plan a real shoot.”
Alex Turner
Music Video Director
“More usable seconds per generation means more material to judge pacing against. Native 4K matters for the same reason — I need detail that survives the review, not just the preview.”
Olivia Martinez
Video Editor
“Changing one variable at a time is the whole discipline. Once I stopped rewriting the entire prompt between attempts I could actually tell what was improving.”
Mohammed Hassan
Digital Artist
“Letting the audio set the timing while images handle the performer and the location is the closest thing to how I plan a real shoot.”
Alex Turner
Music Video Director
“More usable seconds per generation means more material to judge pacing against. Native 4K matters for the same reason — I need detail that survives the review, not just the preview.”
Olivia Martinez
Video Editor
“Changing one variable at a time is the whole discipline. Once I stopped rewriting the entire prompt between attempts I could actually tell what was improving.”
Mohammed Hassan
Digital Artist
“Letting the audio set the timing while images handle the performer and the location is the closest thing to how I plan a real shoot.”
Alex Turner
Music Video Director
“More usable seconds per generation means more material to judge pacing against. Native 4K matters for the same reason — I need detail that survives the review, not just the preview.”
Olivia Martinez
Video Editor
50% off your first annual plan — applied automatically, no codes.
BASIC
Perfect for quick tests and first projects when you are just getting started
1,200 credits/month
Up to 60 videos/month
Up to 240 images/month
Nano Banana Pro
All AI Video Models
AI Music
Private Generation
Ad-Free
Priority Queue
Priority Support
Unlimited Storage
API access (credits shared with web)
Seedance 3.01x Credits
Commercial License
STANDARD
Best for professionals and frequent creators
2,500 credits/month
Up to 125 videos/month
Up to 500 images/month
Nano Banana Pro
All AI Video Models
AI Music
Private Generation
Ad-Free
Priority Queue
Priority Support
Unlimited Storage
API access (credits shared with web)
Seedance 2.5
Seedance 3.01x Credits
Commercial License
PRO
The next step up for serious creators
6,000 credits/month
Up to 300 videos/month
Up to 1,200 images/month
Nano Banana Pro
All AI Video Models
AI Music
Private Generation
Ad-Free
Priority Queue
Priority Support
Unlimited Storage
API access (credits shared with web)
Seedance 2.5
Seedance 3.01x Credits
Commercial License
MAX
For high-volume creators who need maximum credits
13,000 credits/month
Up to 650 videos/month
Up to 2,600 images/month
Nano Banana Pro
All AI Video Models
AI Music
Private Generation
Ad-Free
Priority Queue
Priority Support
Unlimited Storage
API access (credits shared with web)
Seedance 2.5
Seedance 3.01x Credits
Commercial License
Credits are valid for 1 year from purchase. Buy anytime, use anytime.
Starter Pack
Quick top-up for small runs
Up to 70 videos
Includes all features
One-time purchase
No subscription required
Credits valid for 1 year
Creator Pack
Popular for regular usage
Up to 160 videos
Includes all features
One-time purchase
No subscription required
Credits valid for 1 year
Professional Pack
Best for larger batches
Up to 500 videos
Includes all features
One-time purchase
No subscription required
Credits valid for 1 year
Advanced Pack
Best for high-volume usage
Up to 1,650 videos
Includes all features
One-time purchase
No subscription required
Credits valid for 1 year
Power Pack
Best for regular high-volume usage
Up to 3,500 videos
Includes all features
One-time purchase
No subscription required
Credits valid for 1 year
Max Pack
Best for high-volume usage
Up to 9,000 videos
Includes all features
One-time purchase
No subscription required
Credits valid for 1 year
Payment assistance: if you experience any issue during checkout, please contact our support team. support@seedances3.com
How Seedance 3.0 compares to the previous model, what you can create today, and how credits and licensing work.
Seedance 3.0 extends a single generation from 4–15 seconds to up to 30, expands the reference stack from 9 images, 3 videos, and 3 audio files to up to 50 materials, and adds native 4K output. The longer canvas makes full scene planning practical instead of compressing an idea into a short clip.
Have more questions? support@seedances3.com →
Stripe payments
DMCA/CCPA friendly
0+
Used by creators & shops
0+
Videos Generated
Build your reference set, settle your prompt, and generate — everything you need is in the workspace above.