Seedance 2.5NewSeedance 2.0Seedance 2.0 MiniSeedance 1.5 ProSeedance 1.0 ProSeedance ExploreSeedance Blog

Seedance 2.5 AI Video Generatorfor 30s Videos

Seedance 2.5 is live now on JXP. Try it free, no credit card required — upgrade for the full 30-second, 4K experience with more reference control and a cleaner path from prompt to usable clip.

Upload

Click to upload assets

Max 50 files (img ≤30, vid ≤10, audio ≤10)

Duration8s
4s30s

A fast-paced cinematic match-cut short film synced to energetic electronic beats. A flawless crystal sphere stays locked in the exact center of the frame, with a glowing "seedance" logo engraved inside. The sphere remains in razor-sharp focus while the background switches rapidly on each beat: macro water splashes refracting complex light, a vintage morning cafe with steam and commuters, a golden-hour skateboarder tossing the sphere through a rushing street, a music festival with lasers reflected in the glass, a family party table, a dark cinema glow, a vibrating speaker cone that cuts into a DJ turntable, and a warm camping night with firelight bokeh. On the final beat, the sphere flies upward out of frame, cutting to a pure black background with minimalist white "seedance" typography. Dynamic beat-synced editing, cinematic color grading, realistic glass refraction and transmission, ray tracing, global illumination, an extremely sharp subject, strong background motion blur, and high visual impact.

Content Policy Notice

Any non-compliant generation will fail. Due to upstream restrictions, real human faces (selfies, portraits, celebrities) and copyrighted content are not supported. Violent and NSFW content is also rejected. Please use illustrations, anime characters, or AI-generated faces instead.

Overview

What is Seedance 2.5?

Seedance 2.5 is the latest model in the Seedance video family — built for creators who need a finished cut, not a throwaway test. It extends native single-segment output from 15 to 30 seconds with no stitching required to reach the longer runtime, raises the reference ceiling to 50 multimodal inputs — text, images, audio, video, and 3D white-model blockouts — and adds more controllable local editing after generation, so you can fix one detail without re-rendering the full clip.

Key features

More control for AI video creation

Seedance 2.5 matters most when your video idea needs more time, more references, or a cleaner path to revision than a quick clip can provide.

01

30-Second Seamless Output

Generate up to 30 seconds in a single continuous clip - one shot, one prompt, no stitching. Ideal for story beats, product demos, and cinematic sequences that need room to breathe.

Demo 1 / 2

Reference generation

Image × 5
Seedance 2.5 Image × 5 reference 1
1/5
Output
Prompt:One continuous shot. The camera smoothly follows a person in a black coat (reference @Image1) moving left to right through six connected rooms, each with a different tone and atmosphere. Every room shares the same structure: white walls, light herringbone wood floors, French double casement floor-to-ceiling windows, and white sheer curtains, referencing @Image2, but the view outside and the indoor mood are completely different. The protagonist walks at a steady pace throughout, passing through every open doorway in the walls. 0-5 seconds: the first room, themed around American comic-style fighting. The protagonist enters and fights a character (@Image3); the character is defeated. 5-10 seconds: the second room, themed around warmth in a felt-style look. The scene outside the window is a sunflower field (@Image4), with warm orange soft light indoors. A painter is painting sunflowers (@Image5). After the protagonist enters, they also become felt-style. 10-15 seconds: the third room, themed around sadness. The entire image is in black-and-white comic stop-motion style. Rain falls outside; the indoor light is cold, gray, and subdued. A person sits alone on the floor in the center of the empty room, head down, knees drawn up, with a phone beside them showing an unanswered call screen. After the protagonist enters, they turn off the room light, then immediately turn it back on. The room becomes full color and instantly fills with flowers. 15-20 seconds: the fourth room, themed around joy. The entire scene is a room submerged in the sea, referencing @Image6. The protagonist swims into the room, with beautiful coral reefs and schools of fish nearby. 20-25 seconds: the fifth room, themed around surprise. Outside the window is a night sky filled with fireworks, referencing @Image7. Colorful flickering light fills the room. The protagonist is swept into a cheering atmosphere. 25-30 seconds: finally, the protagonist arrives in a blank room, stands in the center, and snaps their fingers. The sound effect is a finger snap. The screen goes black overall, with the word "seedance" appearing in the center, referencing @Image8. Overall cinematic quality, high-end fashion-ad style. Lighting is entirely determined by the scene outside each window, creating strong emotional contrast. No on-screen text.
02

Richer Reference Inputs

Combine up to 50 multimodal references - images, videos, and audio - in one generation. Characters, products, motion, style, and sound can all be anchored before you write the prompt.

Demo 1 / 2

Reference generation

Video x 2
1/2
Image × 13
Seedance 2.5 Image × 13 reference 1
1/13
Audio x 1
Output
Prompt:Core instruction: A full 26-second one-shot narrative short with stable follow-cam, referencing @Video1 and smooth orbiting camera movement, referencing @Video2, interwoven for a smooth sense of forward motion. Day and night alternate and the seasons shift within the same shot. The protagonist is a European woman, @Image1, placed in a crowded, lively street full of human warmth, emphasizing extreme loneliness and cinematic photography quality. Shot breakdown and scene description: 0-3 seconds (steady back follow): An old wooden door, @Image2, creaks open. The camera closely follows the European woman wearing @Image3 as she steps out. She pauses briefly at the threshold. Dappled light fills the lane ahead, and the sounds of hawkers and the crowd rush toward her. She looks detached, then slowly walks into the street. 3-6 seconds (back-side follow): The camera keeps a smooth follow as she enters a crowded morning market, with atmosphere referencing @Video3. Both sides are packed with richly colored fruit stalls and spice shops. A street performance troupe breathes fire, referencing @Image4. Firelight illuminates the crowd, but she never looks aside and walks through at an even pace. 6-9 seconds (smooth side orbit): The camera begins a smooth orbit toward the front-side, capturing the protagonist's profile. She passes a noisy butcher shop, @Image5. A young mother holding a baby, @Image6, brushes past her. The baby stares at her curiously, but she only lowers her eyes slightly to avoid eye contact, without stopping at all. 9-12 seconds (front retreat follow): The camera continues around to the front and follows while moving backward. The crowd ahead suddenly parts naturally to both sides like the Red Sea. A huge elephant draped in ornate red cloth, @Image7, steps in steadily from the right and fills most of the frame. 12-15 seconds (gap pass and return): At the moment the woman and elephant are about to collide, the camera cleverly slides through the narrow gap between them and circles back to her back. The elephant passes by, vast and quiet. Children cheer and chase. Bell sounds and laughter swell, yet she never slows her steps. 15-18 seconds (environmental light shift): As she walks, light within the long take shifts fantastically-the harsh midsummer sun instantly softens. A breeze lifts golden leaves into the air, @Image8, and the season seamlessly transitions to late autumn within the same long shot. Leaves brush her shoulders. 18-21 seconds (360-degree immersive orbit): Ahead, a grand street celebration suddenly erupts, @Image9. Confetti and shredded paper burst into the air. Vendors lean out cheering. The camera now executes a continuous 360-degree orbit, creating an extreme visual contrast between the quiet, lonely protagonist and the frenzied surroundings. 21-24 seconds (return to rear-side): When the orbit completes and returns to her rear-side, the falling confetti has quietly turned into swirling snow-instant winter, @Image10. Passersby open umbrellas or pull up hoods. The woman shivers slightly and raises her coat collar. Her outfit changes to @Image11, and she continues walking alone through the snow. 24-26 seconds (slow push follow): As she walks toward the end of the long street, daylight visibly darkens while she moves. Day seamlessly sinks into night. Warm streetlamps and stall bulbs light up one by one on both sides, @Image12. Vendors pack up their goods. The noise seems slowly absorbed and pulled away by the snow, and her steps gradually slow. Fireworks suddenly bloom in the night sky, @Image13. The sound of the fireworks references @Audio1. Colorful light flickers across building walls and in her eyes. The world remains lively, but she quietly looks up. The camera slowly pulls back and gently concludes.
03

Second-Level Scene Control

Direct pacing, camera language, and emotional beats second by second. Seedance 2.5 keeps scene structure stable while you steer tone, action rhythm, and visual style across the full clip.

Reference generation

Image × 1
Seedance 2.5 Image × 1 reference 1
Output
Prompt:3D animated commercial style, bright and transparent colors. The fruit flesh and juice should feel intensely refreshing and impactful. Overall tone like a high-quality commercial animation short, with a touch of exaggerated humor. The desert horned lizard character is cute, lively, and expressive, referencing @Image1. Image quality should reference the soft natural light, delicate fur / skin texture, dreamy macro depth of field, and a sense of realism with a little playfulness from the reference image. 0-3 seconds: A desert scorched by blazing sun. The air shimmers with heat, the sand burns hot, and the distance looks like it is smoking. A desert horned lizard lies on the scorching sand, tongue slightly out, eyes unfocused, nearly dried out. It takes two steps and wobbles, as if about to "evaporate." Sound effects: heat whoosh, slightly exaggerated cracking dryness. 3-6 seconds: The desert horned lizard suddenly stops and its nose twitches. It looks down and sees a cool, plump grapefruit buried in the sand, covered in water droplets. The grapefruit gleams in the sunlight with delicate skin, like a miracle suddenly appearing in the desert. Performance: the lizard's eyes instantly widen, as if seeing a lifeline. Sound effect: a bright "ding" discovery cue. 6-8 seconds: The desert horned lizard lunges forward and tightly hugs the grapefruit, pressing its whole face against the peel. It shows a blissful "finally alive again" expression. The frame holds for 1 second, creating an exaggerated and funny commercial memory beat. Sound effect: a thud, then half a second of silence. 8-11 seconds: The desert horned lizard grips the grapefruit. The peel splits open, revealing plump flesh with a translucent shine. In the next instant, the juice does not simply flow out-it erupts like a tsunami. Sound effect: a crisp bite crack, followed by an exaggerated juice burst. 11-16 seconds: Orange-pink, clear, glowing grapefruit juice pours out wildly, cascading down the dunes and quickly flooding the entire desert. Dry yellow sand instantly becomes a cool, shimmering, fragrant summer ocean. Cacti, rocks, and small dunes in the desert are swallowed by the juice waves. The image is exaggerated and dreamlike. Performance: the desert horned lizard is excited at first, then realizes something is wrong-its expression shifts from delight to panic. 16-20 seconds: The desert horned lizard is nearly drowned by the "grapefruit sea," desperately clutching half a grapefruit like a life preserver, floating on the surface. It pokes its wet head out with a blank, stunned expression. The sea shimmers, colored like sunlit juice. Sound effects: exaggerated splashing, waves, with a touch of comedy. 20-23 seconds: The screen suddenly cuts to white. Brand text and slogan appear in the center: "Seedance Grapefruit-bite into the flesh, pour out the summer." Voiceover reads the full line: "Seedance Grapefruit-bite into the flesh, pour out the summer." Sound effect: a clean, refreshing brand sting. 23-29 seconds: Cut back from white. The desert horned lizard is now lounging on a floating grapefruit, wearing small sunglasses and holding a cup with a straw, drifting leisurely on the "juice sea" on vacation. Orange pulp, small ice cubes, and cool splashes float around. The sky turns deep blue. The mood shifts from "survival" to "vacation." Finally, the desert horned lizard contentedly leans against the grapefruit. The camera pulls back and settles on a fresh, bright, playful summer image. Sound effects: light summer music, gentle wave lapping. Subtitles may keep only the brand name; no need for extra text.
04

Advanced Controllable Editing

Keep the strongest parts of a generated scene and refine local details - swap outfits, inpaint objects, or restyle environments without restarting from scratch.

Demo 1 / 3

Reference generation

Video x 1
Output
Prompt:Keep the character, jungle environment, camera movement, composition, action rhythm, and duration in @Video1 unchanged. A blue-white energy bow and a glowing arrow slowly appear in the character's hands, @Image1. The bow body gradually forms from faint electric arcs and particles, with delicate flowing current textures, slight volumetric light, and a stable energy outline. During the draw, the arrow condenses at the center of the bowstring into a high-brightness energy arrow. The moment the character releases, the arrow shoots out at high speed, leaving a bright, thin, continuous, sharp energy trail.
05

Expanded Multilingual Presentation

Build one continuous shot where language, typography, and locale shift naturally across regions - ideal for global campaigns, localized intros, and multilingual brand storytelling.

Reference generation

Image × 11
Seedance 2.5 Image × 11 reference 1
1/11
Output
Prompt:One continuous FPV drone first-person video, 30 seconds in a single long take, no cuts, no jump cuts, no transitions. The camera begins inside high-altitude clouds and follows a continuous descending flight path through clouds, mist, light, valleys, waterfalls, lakes, flower fields, city buildings, and a low-altitude plaza. Eleven clearly distinct language-display zones appear in sequence. Each zone shows only one corresponding language, with no mixing, no overlay, and no extra languages. 0-3 seconds: @Image1 clouds naturally form Chinese text reading "Ni hao"; 3-6 seconds: @Image2 mist and volumetric light form English "Hello"; 6-9 seconds: @Image3 high-altitude vapor and sunlight projection form Spanish "Hola"; 9-12 seconds: @Image4 airborne ribbons form Indonesian "Halo"; 12-15 seconds: @Image5 a kite formation forms Malay "Hai"; 15-18 seconds: @Image6 valley morning mist forms Thai text reading "Sawasdee"; 18-21 seconds: @Image7 waterfall mist forms Arabic text reading "Marhaba"; 21-24 seconds: @Image8 lake reflections and water ripples form Portuguese "Ola"; 24-26 seconds: @Image9 flower fields and grass naturally arrange into Vietnamese "Xin chao"; 26-28 seconds: @Image10 city glass reflections and light form Japanese text reading "Konnichiwa"; 28-30 seconds: @Image11 plaza fountain mist, paving, and light strips form Korean text reading "Annyeonghaseyo." Overall morning sunrise atmosphere: golden backlight, soft volumetric light, realistic clouds and mist, natural motion blur, cinematic realism. Camera speed starts slowly at 3-5 m/s, gradually accelerates to 14-16 m/s through natural landscapes, then decelerates to 2-3 m/s for a stable hover over the low-altitude plaza. Camera parameters: wide-angle lens, 24fps, smooth FPV drone movement, pitch gradually transitions from -5 degrees to -18 degrees, then returns to 0 degrees; slight yaw +/-10 degrees, roll controlled at 0-10 degrees, ensuring continuous, stable, realistic one-shot flight feel.
Creations

Made to turn ideas into moving stories

Explore community-style Seedance 2.5 outputs across brand films, surreal scenes, and cinematic storytelling. Hover or tap any creation to play it.

How it works

From brief to video in three steps

Turn a clear brief and the right references into a polished clip, then refine only what needs attention.

Write a Seedance 2.5 video prompt and choose generation settings
01
Step 1

Write the scene prompt

Start with the scene you want Seedance 2.5 to create. Include the subject, setting, camera movement, mood, and the action that should happen across the clip. A longer 30-second result needs a clearer sequence than a five-second test.

Upload visual references for a Seedance 2.5 video
02
Step 2

Add your reference materials

Upload the images, video samples, character references, or brand assets that define the result. With support for 50 reference materials, the strongest Seedance 2.5 workflow uses references to remove guesswork.

Generate and review a Seedance 2.5 video
03
Step 3

Generate, refine, and export

Generate the clip, review consistency, motion, framing, color, and sound across the full scene, then use local editing to fix isolated details without discarding the parts that already work. Export only after checking the final aspect ratio and delivery settings.

Use cases

Built for creators and ambitious teams

Seedance 2.5 fits teams that need a longer AI video draft before they spend time on manual editing, reshoots, or heavy post-production.

For creators making short-form stories
Use case 01

For creators making short-form stories

Read use case details

Seedance 2.5 for creators is useful when a single scene needs time to develop. A 30-second video can carry a full beat: setup, motion, reveal, and ending. That makes Seedance 2.5 a better fit for story-driven clips than tools built only around tiny loops.

For marketers testing product videos
Use case 02

For marketers testing product videos

Read use case details

Seedance 2.5 for marketers can help turn one product idea into a draft ad, demo, or launch teaser. The 50 reference materials support matters here because a product video often needs the item, logo, color palette, target setting, and several style examples to stay on brief.

For agencies preparing client concepts
Use case 03

For agencies preparing client concepts

Read use case details

Agencies can use Seedance 2.5 to show a client a moving concept before committing to a shoot. The value is not replacing every production step. It is giving the client a clearer draft, with enough length to judge pacing and enough edit control to respond to feedback.

For educators and explainers
Use case 04

For educators and explainers

Read use case details

A longer AI-generated clip can help explain a process, demonstrate a scene, or introduce a topic. Seedance 2.5 can be presented as a way to move from static examples to short visual lessons, as long as the final output is reviewed for accuracy.

Compare

See how Seedance 2.5 compares

The clearest comparison is between Seedance 2.5, Seedance 2.0, and general AI video tools that focus on shorter clip generation.

Comparison point
Seedance 2.5
Seedance 2.0
General AI video tools
Single-segment length
30-second single-segment output
Shorter native clip length than Seedance 2.5
Often centered on shorter prompt-to-video clips
Reference input
Supports up to 50 reference materials
Less reference-input emphasis
Usually supports fewer references or simpler prompt inputs
Editing
More controllable for local adjustments
Generation quality is the primary focus
Editing may require regenerating or moving to another editor
Best fit
Longer, more directed video drafts
Established Seedance multimodal workflows
Quick concept tests and short clips

Seedance 2.5 vs Seedance 2.0 is best understood as an upgrade path, not a replacement for every older workflow. Seedance 2.0 remains a strong multimodal baseline, while Seedance 2.5 adds a clearer story around length, reference control, and editable results.

Why choose

More room for better ideas

Choose Seedance 2.5 when the hardest part of your video task is keeping a longer result consistent, directed, and easy to revise.

01

It gives the scene more room

Seedance 2.5 supports 30-second generation, which gives one idea enough time to breathe. You can plan a more complete clip instead of building a timeline from many short fragments.

02

It accepts a richer brief

The 50 reference materials support gives Seedance 2.5 a stronger way to understand the look you want. This is useful when a prompt alone cannot describe the product, character, location, and style.

03

It supports iteration

The video editing direction matters because good AI video work rarely finishes on the first pass. Seedance 2.5 supports a workflow where you generate, inspect, edit, and deliver a cleaner result.

FAQ

Questions, answered

Is Seedance 2.5 out?

Yes — Seedance 2.5 is live and fully available on JXP right now, with no regional rollout limits and no waitlist.

  • Longer generation: Native 30-second single-pass clips, no stitching required.
  • Multimodal references: Up to 50 combined inputs — images, video, audio, and 3D blockouts — in one generation.
  • Local editing: Refine a single detail in the clip without re-rendering the whole scene.
  • Free trial: New users can try Seedance 2.5 free — upgrade to unlock the full 30-second output.
Where can I try Seedance 2.5 online?

You can try Seedance 2.5 directly on JXP — no waitlist, no download required.

  • Free access: New users can try Seedance 2.5 free on JXP — no credit card required.
  • Extended length: Upgrade your account to unlock the full 30-second, higher-resolution generation.
How long can a Seedance 2.5 video be?

Seedance 2.5 supports native single-segment output up to 30 seconds. The full half-minute can be generated as one clip, so you do not need to stitch multiple short generations together just to reach the 30-second runtime.

What is the difference between Seedance 2.5 and Seedance 2.0?

The main Seedance 2.5 vs Seedance 2.0 differences are a longer 30-second native clip, a higher ceiling of 50 multimodal references, and more controllable local editing. Seedance 2.0 remains a strong baseline in the family, while 2.5 is built for longer, more editable production workflows.

Does Seedance 2.5 support text-to-video, image-to-video, and multimodal references?

The multimodal workflow supports text-led and image-led video creation with up to 50 inputs, including text, images, audio, video, and 3D white-model blockouts. These references can anchor characters, products, environments, camera motion, timing, sound, and spatial composition in one brief.

What aspect ratios and video resolutions does Seedance 2.5 support?

JXP currently offers 480p and 720p output for Seedance 2.5, with common aspect ratios such as 16:9 for landscape, 9:16 for TikTok, Reels, or Shorts, and 1:1 for square posts.

Can Seedance 2.5 edit one part of a generated video?

Seedance 2.5 supports local video editing: changing a selected detail, region, or object while preserving the rest of the clip as much as possible. This is intended for fixes such as a product label, hand pose, background prop, or other isolated issue without regenerating the entire video.

How much will Seedance 2.5 cost, and will there be a free plan?

JXP shows Seedance 2.5 credit cost in the creation panel by duration, resolution, and whether reference video is included. Compare the cost of a full 30-second render rather than only a subscription headline.

Will Seedance 2.5 videos have a watermark and allow commercial use?

Watermark removal and commercial-use rights are platform and plan terms, not a capability that should be assumed from the model name alone. Before publishing client or advertising work, confirm the provider’s current license, content policy, attribution rules, and watermark settings.

Will Seedance 2.5 have API access or run locally?

Seedance 2.5 is live for web creation on JXP. Dedicated API access and local deployment are separate from this page workflow—confirm the official model ID, input limits, regional availability, rate limits, and pricing with the provider before production integration.

What makes a good Seedance 2.5 prompt for a 30-second video?

Write a timed sequence with a clear subject, setting, action, camera direction, lighting, and ending. For a 30-second clip, divide the action into a few time ranges, assign each uploaded reference a specific role, state what must remain consistent, and avoid contradictory camera or motion instructions.

Create your next video story

Use Seedance 2.5 when you need an AI video draft with more time, more references, and a better path to revision.

Keep your first project focused: one scene, one outcome, and the exact references that define the result.

Compare Seedance 2.5 features