Seedance 2.5 Is Now Live|Seedance 2.5 brings a more complete video creation experience, offering greater flexibility from idea generation to detail refinement, making creation easier and final production more efficient.
Direct longer AI videos with native sound, precise frames, and up to 20 multimodal references in one production workflow.
Generate from a prompt with native dialogue, music, and sound effects.
Production recipes
These prompt patterns are adapted from the model’s documented workflows: text generation, frame control, multimodal reference, document-to-video, editing, and extension.
20-second cinematic plush-space adventure. A fabric spaceship carrying a rabbit, fox, and crow races past yarn planets while a giant navy-fur whale pursues it. Use exaggerated FPV camera movement, realistic soft-body physics, synchronized impacts, and a triumphant final escape into a pink-violet nebula.
Use Image 1 as the exact opening frame. A swordsman in a dark robe stands in a misty bamboo grove, left hand resting on the hilt as bamboo leaves drift down. One continuous slow push-in; preserve the ink-wash texture and quiet, tense atmosphere.
Replace the person in Video 1 with the character from Image 1. Preserve the original action and camera rhythm, use the voice character from Audio 1, and synchronize the new dialogue naturally with subtle blinking, breathing, and facial emotion.
Turn the reference deck into a premium smart-glasses launch film. Use restrained black, silver, and ice-blue lighting, macro material details, precise parameter graphics, confident urban lifestyle shots, and a clean product-and-tagline end frame.
Edit Video 1 into a refined clay-animation style. Keep the original character identity, action timing, composition, and camera movement unchanged while translating surfaces, lighting, and small details into tactile clay.
Extend Video 1 backward: the baker places the glazed bread into the oven, closes the door, watches it rise, smiles at the aroma, and says, ‘So good.’ Keep the character, kitchen lighting, camera direction, and ambient sound continuous.
Two production modes
Both modes support the complete WAN 3.0 workflow. Standard is the cost-conscious default; Prime prioritizes faster end-to-end delivery.
Best for exploration, story development, and regular production where efficient credit use matters most.
Best for client review, campaigns, and deadline-sensitive iterations where turnaround time matters most.
One directed workflow
WAN 3.0 brings generation, reference control, native sound, editing, and continuation into one consistent production surface.
Generate 2–30 seconds at 30 fps, with timestamped multi-shot direction and native dialogue, music, and sound effects.
Combine images, video, and audio, lock first and last frames, or transform a document or public webpage into a production brief.
Choose adaptive or fixed aspect ratios from 480P to 1080P. Finished videos are copied into NanoPhoto-controlled storage for durable access.