Lippmann's 1908 lens array proposed capturing many viewpoints

↗

A citation that survives in modern engineering literature

Gabriel Lippmann's own paper was not open to read for this draft, but its exact citation and a description of its method both appear in a peer-reviewed source that was: V. Michael Bove Jr.'s 2012 review Display Holography's Digital Second Act, in Proceedings of the IEEE, lists the original as M. G. Lippmann, 'Epreuves reversibles donnant la sensation du relief,' Journal de Physique, volume 4, number 7, pages 821-825, November 1908. Bove's paper explains the method directly: a photosensitive material is placed at the focal plane of a sheet of tiny convex lenslets, each of which captures a different perspective than its neighbours and images it onto a small area of the film. After a replication step that corrects the resulting reversed-depth image, the resulting image is viewed through a lenslet array and recreates the discrete perspectives at appropriate directions, causing the viewer to perceive a three-dimensional image.

Many small cameras standing in for one big one

The optical mechanism is capture and replay through the same kind of lens array. Each lenslet behaves like a tiny separate camera, recording the scene from a slightly different position; the finished print, viewed back through an identical lenslet sheet, sends each lenslet's stored view back out in roughly the direction it was captured from. A viewer moving their head sees the scene shift continuously, because many discrete viewpoints are stored rather than just two fixed ones as in a stereoscope. Viewing conditions follow directly from that: the print must be held at the lenslet array's working distance, and the number of stored views sets how far a viewer can move before the image visibly steps between perspectives rather than flowing.

Why displays still cite a 1908 physics paper

The connection to current work is explicit rather than incidental: Bove's paper treats a modern printed digital hologram as analogous to an integral photograph, with an array of diffractive elements standing in for Lippmann's refractive lenslets. A separately opened source, the Stanford Light Field Rendering paper presented at SIGGRAPH in 1996, treats camera images as two-dimensional slices of a four-dimensional light field to synthesise new viewpoints without depth data, the same many-viewpoints idea in computational form; that paper does not itself mention Lippmann, so the historical link drawn here comes from Bove's account, not from the Stanford paper.

• Is a 'light-field' product capturing and storing many real viewpoints, or interpolating between only a few?
• What working distance and viewing angle was the lens or lenslet array designed for?
• Does a demonstration show continuous motion parallax, or does the image visibly jump between discrete views?

A sheet of tiny lenses recording many viewpoints at once is a genuinely old idea, older than holography itself. What has changed since 1908 is the material doing the recording, not the underlying geometry.