Seedance 2.0
Seedance 2.0 is ByteDance’s multimodal video model. It generates picture and sound together from text, images, audio, and video references, with consistent characters and director-level control.
Built by ByteDance Seed, the team behind the Seedance family
Your generated video will appear here
Multi-reference notes
Add tags such as @Image1, @Video1, or @Audio1 so the model knows which uploaded asset should guide a detail.
Upload limits
- Images: up to 9 files, JPEG / PNG / WebP, 30 MB each, 300-6000px, ratio 0.4-2.5
- Videos: up to 3 files, MP4 / MOV, 2-15s each, 15s total, 50 MB each
- Audio: up to 3 files, WAV / MP3, 2-15s each, 15s total, 15 MB each
- All reference files together: 12 items max and 64 MB total
Use text with image or video references; audio is optional but cannot be used by itself.
What is
What is Seedance 2.0
Seedance 2.0 is ByteDance’s multimodal video model, and it generates picture and sound in a single pass. Feed Seedance 2.0 a line of text, a still image, or a stack of references and it returns a finished clip with motion, audio, and characters that stay consistent from start to finish.
Seedance 2.0 reads up to 12 reference files in one generation: as many as 9 images, 3 videos, and 3 audio tracks. You guide look, motion, and sound at the same time, and Seedance 2.0 keeps them in step.
Native audio is built in. Seedance 2.0 renders sound effects, background music, and lip-synced dialogue next to the picture, so you are not stuck scoring the clip in post.
Consistency is the point. Seedance 2.0 holds a face, a product, and a scene steady across a clip, which is what makes reference-driven work actually hold together.
You direct Seedance 2.0 with a plain @ mention system. Point at @Image1 for a first frame or @Video1 for a camera move, and Seedance 2.0 follows the cue.
What’s new in Seedance 2.0
01 Audio and video in one pass
Seedance 2.0 builds the soundtrack while it builds the picture. Sound effects, background music, and lip-synced dialogue come out aligned with the action, so Seedance 2.0 skips the manual audio pass that earlier video models forced on you. That single change cuts a whole step out of most short-form edits.
02 Twelve references at once
Seedance 2.0 reads up to 12 reference files in a single prompt across images, video, and audio. Hand it a face, a camera move, and a music track together, and Seedance 2.0 weaves all three into one clip. The more you show it, the closer the result lands to what you pictured.
03 Director-level control
Seedance 2.0 gives you a handle on performance, lighting, shadow, and camera movement. A clip looks directed instead of randomly generated, which is why Seedance 2.0 fits real production rather than quick demos. You set the intent, and it keeps the frame on brief.
Built for real production
Native audio
Seedance 2.0 renders sound with the video in one pass: effects, music, and lip-sync arrive already aligned. You skip the separate audio edit that most models leave to you, and the voice tracks the mouth without nudging the timeline. For talking-head and dialogue work, that alone saves an afternoon.
Multimodal input
Text, image, audio, and video all guide Seedance 2.0 in the same generation. Mix them freely to pin down look, motion, and sound at once instead of one at a time. A photo can set the subject, a clip can set the camera, and a track can set the mood, all in one prompt.
Character consistency
Faces, products, and scenes stay consistent across a Seedance 2.0 clip. A character will not drift in appearance or outfit between the first frame and the last, and a product keeps its shape and label. That makes the model practical for brand and catalog work where details matter.
Motion and camera replication
Point Seedance 2.0 at a reference and it copies the choreography, the camera technique, or the editing rhythm. It is handy when you want a known look applied to fresh footage, or when a client sends a clip and asks for "more like this".
Edit and restyle
Seedance 2.0 can replace a character, add or remove an element, or restyle a shot without a full re-render. Small fixes stay small instead of turning into a brand-new generation, so a late note from a reviewer does not cost you the whole clip.
Video extension
Seedance 2.0 extends an existing clip while the motion and story stay coherent, so a short take grows into a longer one without an obvious seam. Stitch a few extensions and a quick idea turns into a full sequence.
Direct with @ mentions
Seedance 2.0 wires references into a prompt with a plain @ mention system. Write @Image1 as the first frame, @Video1 for the camera move, or @Audio1 for the background track, and Seedance 2.0 maps each cue to the right job. Two entry points cover most work: First and Last Frame mode for image-led shots, and Universal Reference mode for multimodal mixes. Either way, Seedance 2.0 keeps the references doing what you told them to do.
A model for every speed and budget
Seedance 2.0 ships as a family rather than a single endpoint. Standard variants cover text to video, image to video, editing, and extension, each with a faster Turbo option for drafts. Seedance 2.0 Mini runs at roughly half the standard price for quick iteration. You match the Seedance 2.0 variant to the job instead of paying full rate for every test render.
More Seedance 2.0 capabilities
Beat-synced editing
Hand Seedance 2.0 a music track and it cuts the visuals to the beat, music-video style, without manual trimming. The edit lands on the rhythm you gave it, so transitions hit on the downbeat instead of drifting a frame off. It is a quick way to make a plain clip feel produced.
One-take continuity
Seedance 2.0 can hold a long, unbroken shot with steady motion, so a single take reads as one continuous move rather than a stitched sequence. The camera glides, the subject holds, and nothing pops between cuts because there are no cuts to begin with.
Timeline prompting
Break a prompt into time windows, like 0 to 2 seconds and 2 to 4 seconds, and Seedance 2.0 follows the beats in order. You script the clip moment by moment, which is the difference between a lucky generation and a planned one.
How to make a clip with Seedance 2.0
Upload your references
Drop up to 12 files into Seedance 2.0: as many as 9 images, 3 videos, and 3 audio tracks. These anchor the look, the motion, and the sound of the shot. Cleaner references give the model less to guess at, so it is worth picking sharp ones.
Write the prompt with @ mentions
Describe the shot in plain language and point at your assets with @ tags. Seedance 2.0 reads @Image1, @Video1, and @Audio1 as the first frame, the camera reference, and the soundtrack. Spell out who does what, and the model has fewer ways to wander off brief.
Pick mode, length, and ratio
Choose First and Last Frame or Universal Reference, set a length from 4 to 15 seconds, and a ratio like 16:9 or 9:16. Seedance 2.0 adapts to the format you pick, so a vertical ad and a landscape trailer come from the same prompt.
Generate, then edit or extend
Seedance 2.0 returns the video with audio in one pass. From there you can re-edit a region or extend the clip without starting from scratch, which is how a first draft turns into a finished cut in a few rounds.
Seedance 2.0 vs Seedance 2.5
Clip length
Seedance 2.0 runs 4 to 15 seconds per generation. Seedance 2.5 stretches a single continuous shot to 30 seconds.
References
Seedance 2.0 takes up to 12 reference files. Seedance 2.5 reads as many as 50 at once.
Resolution
Seedance 2.0 outputs up to 1080p. Seedance 2.5 moves up to native 4K with 10-bit color.
Editing
Seedance 2.0 edits and extends clips. Seedance 2.5 adds region-level re-draw and 3D blockouts on top.
What people make with Seedance 2.0
Seedance 2.0 FAQ
What is Seedance 2.0?
Seedance 2.0 is ByteDance’s multimodal video model. It generates video and audio together from text, images, audio, and video references, with consistent characters and director-level control.
Does Seedance 2.0 generate audio?
Yes. Seedance 2.0 renders native sound effects, music, and lip-synced dialogue in the same pass as the picture, so you do not add a soundtrack afterward.
How long are Seedance 2.0 clips?
You can set a length from 4 to 15 seconds per generation. For longer pieces, Seedance 2.0 can extend a clip or chain takes while the motion and story stay coherent.
How many references does Seedance 2.0 take?
Up to 12 files in one generation: as many as 9 images, 3 videos, and 3 audio tracks. You wire each one into the prompt with an @ tag, so Seedance 2.0 knows which asset does which job.
What resolution does Seedance 2.0 output?
Seedance 2.0 outputs up to 1080p, across ratios like 16:9, 9:16, and 1:1.
Can Seedance 2.0 edit or extend a clip?
Yes. Seedance 2.0 can replace a character, add or remove elements, restyle a shot, or extend footage while the story stays coherent.
What is the @ mention system?
It is how Seedance 2.0 wires references into a prompt. Tag @Image1, @Video1, or @Audio1 and Seedance 2.0 maps each to the first frame, a camera move, or the soundtrack.
Does Seedance 2.0 have a faster or cheaper option?
Yes. Turbo variants cut latency, and Seedance 2.0 Mini runs at about half the standard price for quick drafts.
What can I build with Seedance 2.0?
Social clips, product ads, music-synced edits, storyboards turned into video, and localized content all suit Seedance 2.0. Because it renders sound with the picture and holds characters steady, it fits both quick one-off posts and longer branded pieces that need to stay on message.
How is Seedance 2.0 different from Seedance 2.5?
Seedance 2.0 runs 4 to 15 seconds at up to 1080p. Seedance 2.5 extends a single shot to 30 seconds, reads up to 50 references, and outputs native 4K. For short, audio-rich clips, Seedance 2.0 covers most of the work.
Does Seedance 2.0 do lip-sync?
Yes. Seedance 2.0 syncs spoken dialogue to the mouth as part of the same render, so a character can talk on camera without a separate lip-sync step or audio cleanup afterward.
Is Seedance 2.0 good for vertical social video?
It is. Seedance 2.0 supports 9:16 along with 16:9 and 1:1, and the native audio means a TikTok or Reels clip lands with sound and motion already in place.
Make your first Seedance 2.0 clip
Type a prompt or drop in an image. You get cinematic video with sound, straight from the browser.
Try Seedance 2.0