seedance-core

To write Seedance video prompts from a reference image, you need to organize scene concepts into a master prompt structure covering subject, wardrobe, environment, style anchor, delivery, logic rules, negative prompts, and action.

Multi-cut B-roll and single continuous shots require different prompt structures. You must distinguish between shot families during prompt generation to ensure the output matches the intended edit structure.

Audio timing drives lip sync video prompting by enforcing precise synchronization between dialogue and visual delivery. Decisive prompt choices and audio-driven timing rules prevent failed generations in talking head shots.

Reference scoping defines visual boundaries before Higgsfield video generation. It enforces reliability guardrails by verifying start frames and applying negative prompt discipline, reducing wasted paid generations.

Limitations in montage sequence prompt craft include failing to make decisive choices between single continuous shots and multi-cut B-roll, which causes mismatched edit structures and unreliable generation outputs.