Visual Prompt Reverse Engineering

Reconstruct a reference image as a reproducible visual specification, then express that same specification as aligned Chinese and English generation prompts. Describe what is visually supported; do not invent hidden facts, identities, brands, locations, stories, or camera metadata.

Non-negotiable invariant

Maintain semantic equality throughout the workflow:

Visual Facts (Chinese) = Visual Facts (English) = Master Visual Specification

“Equal” means the same visual constraints are present at the same specificity and certainty. Natural wording and order may differ. Neither language may add, omit, weaken, strengthen, or contradict a fact.

Required pipeline

Run these stages in order. Do not draft the English prompt early.