Details
Photographers activate AI Masks by clicking on part of an image, drawing a bounding box, or selecting from predefined subject categories (sky, face, hair, background, people, animals). The machine learning model produces pixel-level selections that photographers then use to apply local corrections such as exposure, noise reduction, or color adjustments only to the selected area. Version 9.2 (released November 2025) improved edge precision and sensitivity thresholds; version 9.6 (March 2026) added feathering controls. All AI processing occurs on the user's local machine, without uploading images to external servers.
Have evidence about DxO's AI practices? Submit a report.
Report a Sighting →