Map reference images into selectable named elements, produce evidence-labeled reconstruction JSON, and compile it into prompts for reconstruction, remixing, and scoped edits. Use when a person wants to understand, recreate, or selectively change an image.
Langue du texte source : anglais