Depth difference between left and right camera image in stereography — larger disparity makes objects appear closer. Core parameter for 3D convergence.
On the set of a stereoscopic production, the question arises sooner or later: How close should the object really appear? The answer lies in disparity—the pixel offset between the left and right camera images. The greater this offset, the closer the subject appears to the viewer; the smaller, the further away. Disparity is the measurement tool for spatial depth in stereoscopic capture.
In practice, it works like this: You set up two cameras with a known distance (interaxial distance). Both look at the same point, but from slightly different positions. A nearby object creates a large offset between the left and right image—high disparity. An object on the horizon appears almost identical in both images—low disparity, practically zero. The viewer, with their 3D glasses, receives this offset in their brain and reconstructs depth from it, just like in biological vision.
The critical parameter is the convergence line—the point to which both cameras are geometrically aligned. Everything in front of it has positive disparity (bulges towards the viewer), everything behind it has negative disparity (recedes into space). In post-production, especially during depth mapping or digital stereo conversion, disparity becomes a control variable: you determine how far forward certain objects should appear by manipulating their disparity values. An incorrect value—and the depth collapses, giving the viewer a headache.
On set itself, you must control disparity: too large an interaxial distance with nearby subjects creates uncontrolled "floating" and cross-eye effects. Too small a distance with distant objects provides no depth. The trick is to find the sweet spot—usually between 6 and 8 cm for normal dramatic shots, up to 30 cm for landscapes. On the matte painting side or in VFX shots, disparity is then set manually: the compositor determines the depth of each layer using a Z-depth pass. This requires precise control because errors are immediately visible if depth and motion don't align.