Face Clustering Engine
A reviewable visual-clustering system for finding recurring synthetic characters across changes in pose, style, and lighting.
Synthetic-character archives accumulate near-matches and visual changes that ordinary folders cannot organise reliably.
Indexed portraits feed reusable identity representations and proposed visual groups. The review surface keeps the current members and neighbouring examples close at hand, so ambiguous group boundaries remain inspectable.
Groups are proposals, not identity verdicts. An operator can correct a tentative assignment, undo a recent decision, and separate uncertain or unusable material instead of forcing every image into a label.
That keeps the archive navigable while preserving a human route through ambiguity and recovery.