Guide · Video
Seedance 2 is one of the strongest video models inside Plenty Labs. The difference between a random clip and a cinematic one is almost never the model. It is the structure of the prompt. This guide gives you the structure.
Describe one scene per prompt, in this order: subject, action, setting, camera, light, mood. Short concrete clauses beat long poetic ones. Instead of a beautiful cinematic scene of a woman in a city, write: a woman in a grey wool coat crosses an empty crosswalk, early morning, slow push-in at eye level, soft overcast light, calm and unhurried.
Decide aspect ratio and duration first, then write for that frame. Vertical 9:16 wants tighter framing and closer subjects than 16:9. Keeping one locked format convention across your team makes every prompt reusable and every output consistent.
Seedance 2 responds strongly to camera language. Name one movement per shot: static tripod, slow push-in, handheld follow, orbit. Stacking multiple camera moves in one prompt is the most common reason a clip feels chaotic. One shot, one move.
When a render is close but not right, change exactly one thing: the light, the lens, or the action. Changing three things at once teaches you nothing. When a prompt hits, save it as a preset in Plenty Labs so the whole team can start from it next time.
