Strictly speaking, this is not a philosophy of how to determine the convergence point, but one of "linking" convergence with interaxial so that the relationship between them is a constant. Propounded by Paul Vlahos, it was used with some modifications by Ken Jones in filming some of the effects for Jaws 3-D.
Convergence can be done during photography or in a later optical stage with the help of a reference target. The basic philosophy is that a ratio is established and maintained between interaxial and convergence distance. As the point of convergence is brought closer, the interaxial is reduced in direct proportion. Keeping the same amount of toe-in between the cameras (or lateral shift of the two images), a change in interaxial will automatically change the convergence to a specific distance. The convergence distance is selected and a scale gives the setting for interaxial.
With this technique and a constant focal length, infinity points remain at a constant (apparent) distance from the audience and if the main subject is converged upon, it too remains a constant distance from the audience. Infinity points can be "locked in" with the choice of the interaxial-convergence ratio, so they need never go beyond comfortable limits. The closer the subject, the closer the convergence will (probably) be placed and the narrower the rnteraxial becomes, preventing excess distortion of closeups due to toowide an interaxial.
Limitations this philosophy place upon photography are partially aesthetic and partially practical:
1. The choice of convergence always determines the maximum possible scene depth; the same convergence will always result in the same maximum possible depth. This can hamper the versatility of the 3D filming. Also, the closer the convergence is placed, the flatterthe scene and everything in it becomes - just the opposite of that normally occurs when we look at an object closer to us.
2. If a different focal length is used, a new .ratio of interaxial to convergence may have to be used; ideally there should be a different ratio for each focal length.
3. If the cameras are converged ("toed-in") during photography, the camera rig must be capable of maintaining the same angle of toe-in while interaxial is changed; if this involves moving the toed-in camera(s), this could require very tricky engineering of the rig. If the convergence function is to be done in a later stage, this requires an overall optical correction in the lab. While this can be easily done, it is an extra step in which something can go wrong.
4. This philosophy can only be used with 3D systems which have variable interaxial capability over a sufficient range. This restricts its use almost exclusively to double-camera systems.
Maximum Near Object Convergence
This seems to be the preferred philosophy by amateur stereo photographers shooting stills. Stereo slides for local, national, and international competitions frequently abide by this philosophy. Although an unusual one for 3D cinematography, there is no a priori reason why it cannot be applied to motion pictures.
The technique is simply to converge on the nearest object in the scene, at whatever distance from the camera it is. The result is that the nearest object(s) will be at the plane of the screen and everything else in the scene will be behind the screen. The effect will be one of looking through a window at the 3D scene.
This philosophy has the great advantage of eliminating any problem of objects being too far in front of convergence, and either looking ambiguous because they are "cut off" by the edges of the picture or protruding so far off the screen that they are uncomfortable or impossible to look at stereoscopically by the audience. Additionally, the eyes of the spectators are required to converge only from the screen back to (an assumed) comfortable infinity; this reduces the amount of angular movement of the eyes away from their focus point, reducing fatigue. …