30-second directed clips
The extended ladder (up to 30 seconds) holds true three-act mini-films with setup, escalation, and payoff in one generation.
Seedance 2.5 is an AI video generation model developed by ByteDance. Use Seedance 2.5 as the flagship AI video generator for clips up to 30 seconds. Learn its frame control, reference system, prompting, examples, and limits before generating. The sections below break down its capabilities, live settings, prompting, examples, and review points.

MODEL OVERVIEW
Seedance 2.5 is the flagship: longest duration, deepest references, and explicit frame control for directed mini-films.
Try Seedance 2.5
2.5 targets directed work: up to 30 seconds at up to 1080p with first/last-frame bracketing and reference audio and video.
Text, image, and reference-to-video modes accept up to 30 references with 30,000-character briefs for genuinely specified productions.
Use the live settings table as truth for the extended duration ladder, resolutions, adaptive ratio, and advanced generation types.
CORE CAPABILITIES
Why flagship productions pick 2.5.
The extended ladder (up to 30 seconds) holds true three-act mini-films with setup, escalation, and payoff in one generation.
Bracket motion with explicit start and end states so logo resolves and transformation payoffs land exactly as boarded.
Feed performances and tracks as direction: acting reference drives motion while audio reference drives timing.
Cast, wardrobe, locations, and style frames combine into one controlled world when every input carries a labeled role.
LIVE SETTINGS
These values come from the currently configured 2.5 route, so this table stays aligned with what the generator can actually submit.
HOW TO USE IT
A director's loop for flagship clips.
Storyboard beats, cast references, and the timing map before spending flagship credits.
Supply first and last frames so the generation bridges exactly your endpoints.
Upload cast, set, and style inputs with labeled roles plus any reference audio or video.
Check performance, sync, identity, and design fidelity at full resolution before finaling.
BEST USE CASES
Pick 2.5 when the piece has direction, not just motion.
2.5 carries 30-second brand films and lore pieces with composed openings and closings.
Reference video transfers real acting nuance that text prompts approximate poorly.
Score, dialogue, and effects lock to picture with frame-level intent from reference audio.
Thirty references sustain recurring casts and locations across episodes without drift.
PROMPT GUIDE
Long, specified briefs worthy of the flagship.
Try Seedance 2.5With up to 30,000 characters, specify blocking, lens, light, and performance instead of compressing intent.
Let reference video carry acting and reference audio carry timing; text then only needs the gaps.
Mark identity, logo, and design elements as locked so the flagship protects them over unlisted details.
PROMPT EXAMPLES
Starting points for directed productions.
A fencer advances down a sunlit piste, first frame en garde, last frame victory salute, reference footwork video attached, 20 seconds.
Noodle chef pulls dough in a steamy kitchen, flour bursts on the drum hit from the reference track, ends on the served bowl close-up, 25 seconds.
Two astronauts plant a flag on a ridge as the lander glints behind, opening wide, closing on the patch detail, 30 seconds, 1080p.
WHAT TO REVIEW
Flagship power with flagship costs and complexity.

Highest per-clip cost and longest waits in the Seedance line; rehearse direction on cheaper tiers first.
Thirty references demand real art direction: unlabeled or conflicting inputs produce expensive confusion.
Extreme choreography (crowds, combat, intricate hand work) still needs review at full resolution before delivery.