30-second storytelling
Generate up to 30 seconds in one pass with improved transitions, scene changes, audiovisual continuity, and support for extending the result.
Learn how Seedance 2.5 approaches 30-second audio-video storytelling, flexible multimodal references, and precise editing, then explore the current Seedance 2.0 workspace while the 2.5 route is being prepared.
Seedance 2.5 is not yet connected to Nano Banana Pro. The workspace currently opens Seedance 2.0 by default; model options, duration, points, and submitted tasks belong to the selected live model, not Seedance 2.5.
Generated content will appear here
Seedance 2.5 is not yet connected to Nano Banana Pro. The workspace currently opens Seedance 2.0 by default; model options, duration, points, and submitted tasks belong to the selected live model, not Seedance 2.5.
Facts reviewed:
Seedance 2.5 is ByteDance Seed’s next-generation audio-video creation model, officially announced on July 31, 2026. It builds on the unified multimodal audio-video architecture introduced with Seedance 2.0 and focuses on longer storytelling, stronger reference understanding, and more precise generation and editing controls.
According to ByteDance Seed, one generation can produce a clip up to 30 seconds and can be extended for longer narratives. The model can accept up to 30 images, 10 video clips, and 10 audio clips as references in one request, using them to understand subjects, scenes, composition, movement, camera language, style, and sound. Those are official model capabilities, not the controls currently exposed by Nano Banana Pro.
Nano Banana Pro has not yet connected a Seedance 2.5 task route. This landing page provides a fact-checked model guide and a temporary Seedance 2.0 workspace so creators can prepare prompts and understand the existing video interface. Always check the model name and point estimate shown inside the workspace before submitting.
The official release focuses on complete stories, richer references, and production-oriented control.
Generate up to 30 seconds in one pass with improved transitions, scene changes, audiovisual continuity, and support for extending the result.
Combine images, video clips, and audio clips so the model can follow characters, props, composition, motion, style, camera language, voices, and sound.
Plan visuals and sound together, including dialogue, ambience, effects, music, timing, and how audio supports the narrative.
Write prompts in timed segments to control when actions, camera changes, dialogue, and story beats happen within a clip.
Use targeted instructions to revise characters, actions, shots, or plot details while aiming to preserve continuity around the edited segment.
Official examples include clay-model or white-model control, green-screen editing, camera-perspective editing, blocking, and complex production workflows.
The official release focuses on complete stories, richer references, and production-oriented control.
Plan longer product stories, camera blocking, sound cues, and transitions before committing to a physical shoot.
Explore scenes with multiple story beats, characters, locations, dialogue, and audiovisual continuity in one generated sequence.
Use approved brand images, motion references, audio, and style material to communicate a more specific creative direction.
Turn processes, historical scenes, scientific concepts, or equipment workflows into structured audiovisual explanations.
Structure mobile-first scenes around wardrobe continuity, controlled movement, texture, and sound rather than a single static pose.
Plan repeated subjects, props, travel direction, and audio bridges across several environments or times of day.
These prompts are structured for the newer model, but the buttons currently load them into the temporary Seedance 2.0 workspace for editing.
Organize a continuous commercial scene into timed beats without relying on abrupt cuts.
16:9 cinematic one-take product story. 0–5s: macro view of a ceramic tea cup as morning steam catches warm window light. 6–10s: the camera pulls back and follows a craftsperson carrying the cup through a quiet studio. 11–15s: circle around the table as ingredients and tools come into view. Keep the same cup, hands, lighting direction, and spatial continuity. Natural room ambience, subtle ceramic sounds, no subtitles or logos.
Use an authorized image as the visual anchor and describe only the intended motion and sound changes.
Preserve the identity, clothing, color palette, and environment from the uploaded image. The subject turns toward the window, takes two natural steps, and gently opens the curtain as the camera makes a slow lateral move. Maintain facial structure and room geometry. Add soft fabric movement, distant city ambience, and a quiet breath; avoid cuts, extra people, and text.
Align performance, camera, and audio cues by giving each section a clear time window.
Naturalistic short drama, consistent characters and location. 0–5s: medium two-shot in a rainy station café; one character says softly, “You came back.” 6–10s: slow push toward the second character as they answer, “Only for tonight.” 11–15s: hold the silence while a train passes outside and window reflections move across their faces. Accurate lip sync, restrained acting, rain and train ambience, no background music.
Plan a vertical editorial scene that protects wardrobe, identity, and material details across timed beats.
9:16 fashion editorial in one continuous take. 0–4s: full-body portrait beneath a rain-lit arcade, silver coat and black boots clearly visible. 5–9s: the subject walks toward camera while reflections ripple across the floor. 10–15s: camera passes beside the subject and reveals a neon street beyond. Preserve face, garment construction, accessories, and walking direction; restrained fabric motion, rain ambience, distant traffic, no captions or brand marks.
Use time blocks, repeated props, and sound bridges to hold a story together across an environment change.
Cinematic narrative with one recurring red bicycle. 0–5s: sunrise on a quiet coastal lane as a rider enters frame. 6–10s: match the wheel rotation into a crowded afternoon market, keeping rider, bicycle geometry, and travel direction consistent. 11–15s: emerge onto a blue-hour bridge and settle into a wide rear view. Natural wheel and chain sound, market ambience blending into river wind, no logos, no abrupt cuts.
Write the setup, development, turning point, and ending before describing individual visual details.
Use only authorized images, videos, and audio, and state exactly what each reference should control.
Break the prompt into time windows and align actions, camera, dialogue, ambience, and transitions.
Before generating, confirm that the workspace actually shows Seedance 2.5 and review its current parameters and point estimate.
Seedance 2.5 is not yet available as a Nano Banana Pro task route. The embedded workspace currently selects Seedance 2.0, which has different duration, reference, quality, and pricing controls. A prompt loaded from this page does not change the selected model into Seedance 2.5.
ByteDance Seed notes that physical plausibility in complex motion and stability in scenes with multiple interacting subjects can still improve. Generated people, hands, objects, subtitles, dialogue, timing, and sound may be inaccurate. Review the entire video and soundtrack before publication or client delivery.
Only upload material you are authorized to use. Do not use generated video for deceptive impersonation, harassment, privacy violations, intellectual-property infringement, or false evidence. Model access, API support, quotas, safety rules, and pricing can change; the official ByteDance source and the live Nano Banana Pro interface remain authoritative for their respective services.
Model capabilities come from official ByteDance Seed material. Nano Banana Pro integration status is stated separately and was reviewed on the date shown above.
Use the current workspace to refine a prompt, but verify the selected model before generating. Seedance 2.5 availability will be updated after the route is connected.