skills/seedance-2.0-prompter/references/atomic_element_mapping.md

Atomic Element Mapping Knowledge Base

This document provides the core logic for the Seedance Prompt Designer skill. It contains two critical mapping tables.

Table 1: Asset Type -> Potential Atomic Elements

This table maps the type of user-uploaded asset to the most likely atomic element roles it can play in video generation. The skill should use this table in Phase 1 to analyze uploaded assets.

Asset Type (Heuristic)	Potential Atomic Element(s)
Image with a clear human/character face	Subject Identity, Aesthetic Style
Image of an object/product	Subject Identity (Object)
Image of a landscape/environment	Scene Environment, Aesthetic Style
Image with strong artistic style (e.g., painting, sketch)	Aesthetic Style
Image with clear compositional structure	Composition / Layout
Video with a character performing an action	Subject Motion, Subject Identity
Video with significant camera movement	Camera Language
Video with prominent visual effects	Visual Effects
Audio file with speech	Voice (Timbre)
Audio file with music	Non-speech (Music)
Audio file with sound effects	Non-speech (Sound Effect)

Table 2: Atomic Element -> Optimal Reference Method

This table defines the best way to reference each atomic element when constructing the prompt. The skill must use this table in Phase 2 to design the reference strategy.

Atomic Element	Optimal Method	Rationale
Subject Identity	Asset	High information density. Must use a reference image.
Scene Environment	Hybrid	Use an asset for the base, and text to modify details (e.g., weather).
Aesthetic Style	Hybrid	Use an asset to define the style, and text to specify its application.
Composition / Layout	Asset	Purely visual. Must be controlled by a keyframe image.
Subject Motion	Hybrid	Simple, describable actions use text. Complex, unique motions require a reference video.
Camera Language	Text	Standardized cinematic language. Text is clearer and more direct.
Visual Effects	Text	Most effects are describable. Only very unique effects need a reference video.
Voice (Timbre)	Asset	Unique biometric signature. Must use an audio sample.
Voice (Performance)	Text	Performance details (speed, tone, emotion) are best controlled by text/SSML.
Non-speech (SFX / Music)	Asset	Unique sounds/melodies must be provided as an asset.
Multi-shot	Text	Structural information is best defined by structured text (e.g., multi_prompt).
Pacing	Text	Temporal control is best defined by text parameters.
Story Logic	Text	Abstract concepts can only be guided by text prompts.