Two synchronized cameras with calibrated inter-ocular distance shooting simultaneously for stereoscopic image — industry standard since Avatar. Demands precise calibration and timecode sync.
Two cameras, perfectly synchronized, with identical focal lengths and a calibrated interpupillary distance between the sensors—this is the foundation of modern 3D blockbuster production. While you're working on set with this setup, one camera controls the left eye, the other the right. The distance between them isn't arbitrary: it corresponds to human binocular vision but can be adjusted depending on the shooting distance and desired stereoscopic depth. For closer subjects, you reduce the interaxial distance to avoid over-convergence; for greater distances, you increase it for more spatial impact.
The central problem: synchronization. Both cameras must run identically on the exact frame—timecode lock is not optional. In editing, this method only works if your takes are pixel-perfectly aligned. A frame offset of two images and the entire sequence becomes unwatchable; the viewer's eye cannot fuse. Therefore, on set, you need a 3D supervisor who constantly monitors the interaxial setup, sets the convergence plane, and physically synchronizes both cameras. Most modern cinema cameras offer trigger sync—one becomes the master, the other follows electronically. But even this requires regular calibration and test takes before shooting.
In practice, the dual-camera rig has established itself as the standard since Avatar because it delivers true parallax—no post-production, no monocular 3D tricks that trick the eye into seeing what isn't there. You see the spatial information directly in the live image on the monitors. The disadvantage: the rig becomes cumbersome. Steadicam work becomes complex, crane moves require special rigs, and any focal length adjustment must be done precisely on both cameras. Handheld is theoretically possible, but the operator must work with extreme discipline—minimal vibrations, minimal relative movement between the cameras.
In editing, you then work with two image sequences in parallel—both must run on the timeline, both are merged into a stereo pair. Errors here are irreparable. Therefore, the AC and editing assistant log every take with the interaxial distance, the convergence plane, and exact sync data. Without this metadata, stereo editing becomes a blind flight. The DCP distribution then occurs as a stereo file, read by the cinema projector as the left and right eye.