The Complete Guide to Seedance 2.0 Cinematic Prompting
Most AI video prompts fail for one reason: they describe what is in the shot but not how it's filmed. The difference between stock footage and cinema is almost never the subject — it's the language of the camera around it.
This guide breaks down the seven-block structure that consistently produces cinematic results across the Seedance 2.0 collection.
The seven blocks of a cinematic prompt
A strong prompt reads like a shot list. Each block answers one question a cinematographer would ask on set.
- Shot type — how close are we? Sets emotional distance before anything else.
- Subject and action — what happens, in concrete verbs.
- Setting — where, with texture.
- Camera movement — static, push-in, orbit, crane, FPV. Where a shot starts to feel directed.
- Lens — 35mm, 85mm, anamorphic, macro. Quietly controls intimacy and falloff.
- Lighting and mood — the single biggest lever on atmosphere.
- Color grade — what makes the eye read "cinema."
Put together, the blocks compound:
Cinematic wide shot of a lighthouse keeper releasing paper lanterns into a storm.
Camera: slow push-in, anamorphic lens. Neon glow lighting. Dreamlike atmosphere. Teal and orange color grade.
Remove the camera line and it deflates; remove the grade and it stops looking like film.
Why structure beats inspiration
Freehand prompting can't be debugged — when it works you don't know why, when it fails you don't know which part to fix. Treating the prompt as fixed slots turns prompting into something you can debug. Too flat? Change the movement. Too cold? Change the grade. You stop guessing, and your clips stay coherent across a project because they share a grade and a lens language.
Matching structure to genre
The seven blocks stay constant; the emphasis shifts. Horror leans on lighting and the unseen. Nature documentary leans on lens and patient movement. Music video leans on rhythm and grade. Learning which blocks carry the weight in each genre is most of the craft — which is why a library organized by genre beats a flat list.
Format is part of the prompt
A prompt built for 16:9 and one for 9:16 are not the same shot — framing, headroom and camera distance all change. Append the format explicitly ("Vertical 9:16 format") so the model composes for the frame you'll publish in.
See the structure in action
The fastest way to internalize this is to build a few shots and watch the structure assemble. Once it clicks, you'll never write "a woman walking through a city" again.