Read the explanation
This source is a deliberate prompt checklist, not an image generator or a model benchmark. It counts three possible considerations: demographic context, cultural or geographic setting, and casting intent without stereotypes. Selecting two shows two out of three checks considered; selecting all shows three out of three. Both bars use one hundred fifty pixels per selected check. This count measures which boxes the author considered, not representational quality or fairness of a generated result. Editing the scene text updates one source state field named prompt. Each checkbox change rebuilds the checks array from the checked inputs. The render function displays the count, then saves the prompt and checks to browser local storage. This diagram follows the actual source data path. No request is sent to a generation model, and no identities are inferred from images. A fresh browser context and a persistent reload can test the storage behavior without using real user data. The Export review button writes the exact prompt and selected checks to a JSON file. It also includes the source review-mode and boundary strings stating that no image is generated, no identity is inferred, and no model benchmark is claimed. Reset restores the original prompt and an empty checks array. The source limitation remains visible even when all three considerations are selected. The export records an authorship review; it does not certify a future model output.