Subject Recognition AF

Subject recognition autofocus uses machine learning models on the camera's image processor to identify specific subjects in the frame and continuously adjust focus to keep them sharp. Rather than focusing on the nearest object or wherever a focus point is manually placed, the system detects the category of subject being photographed, locates it in the frame, and tracks it as it moves. Subject categories available on M43 cameras include humans with face and eye detection, birds, animals, and vehicles.

Eye detection is the most precise form of subject recognition for portrait and people photography. When a human subject is detected, the camera identifies and locks onto the nearest eye, maintaining focus at the eye level even when the subject moves or turns. On M43 cameras with PDAF, this works reliably in continuous AF mode during both stills bursts and video recording. The phase-detection data provides the real-time distance information needed to keep the eye in focus as the subject moves toward or away from the camera.

Bird detection specifically identifies birds with beak, eye, and body recognition, maintaining focus during erratic flight movement. This was introduced to the M43 system through firmware updates on several OM System and Olympus bodies, including the E-M1X and E-M1 Mark III. Panasonic introduced equivalent animal and bird recognition on the G9 II and GH7. Subject recognition performance continues to improve with firmware updates on existing bodies.

Subject recognition AF is most practically useful for wildlife photography and portrait work with fast primes at wide apertures. For wildlife, it removes the need to manually reposition a focus point as a bird or animal moves across the frame. For portraits with shallow depth of field, eye detection removes the need to aim a focus point precisely at the subject's eye, which is critical when depth of field at f/1.2 or f/1.4 is measured in centimetres.