One panel, many small images
Integral imaging puts a sheet of tiny lenses, a microlens or lenslet array, directly over a flat panel or, in the original photographic version, over film. Each lens captures or displays a small 'element image': a miniature view of the scene from that lens's slightly different position. Enough element images side by side reconstruct both horizontal and vertical parallax, full parallax rather than the single left-right axis a parallax barrier or lenticular strip provides. A 1997 Applied Optics paper by Okano, Hoshino, Arai and Yuyama replaced the original film-based capture with a television camera photographing the array of real images directly, enabling moving pictures rather than the single stills conventional integral photography had been limited to.
What creates the image, and what it takes to view it
On replay, the same kind of lens array sits over a display panel; each lenslet re-emits its element image's rays back out at the corresponding angles, so a viewer's two eyes intercept two of the many angular samples and perceive real depth, no glasses required, with no single fixed sweet spot the way a two-view barrier display has. The trade-off is resolution: the panel's pixels are divided among however many element images the array captures, so a denser array giving smoother parallax means a coarser image, and a sharper image means fewer viewing angles. A follow-up 1998 paper by Arai, Okano, Hoshino and Yuyama names two specific defects this trade-off produces: pseudoscopic, depth-reversed images, and interference between neighbouring element images, both of which their gradient-index lens design was built to correct.
Where the survey literature places this approach
Neither Applied Optics paper opened here uses the phrase 'integral imaging' interchangeably with terms like light-field or multiview display, and a broader 2007 3DTV display survey categorises full-parallax multiview systems as a class without naming integral photography specifically in its abstract. That gap is worth flagging rather than smoothing over: this record establishes what the two founding papers demonstrate about lens-array capture and replay, not how every later commercial 'integral imaging' or 'light-field' product compares to them. Treating the two labels as identical is an editorial simplification some vendors rely on.
Questions to bring to a demo
- How many element images does the array actually capture per resolvable point, and what spatial resolution results?
- Is vertical parallax present, or only horizontal, which would suggest a lenticular rather than a true lens-array system?
- How has pseudoscopic reversal or inter-image interference been corrected, if at all?
A lens array buys full parallax at a real, measurable resolution cost; the two 1990s papers that built the method also documented its specific failure modes rather than only its promise.
Sources & provenance
Preserved from the existing project research collection. These records were not independently re-researched as part of the design update. Supplier, historical and technical claims remain subject to editorial review; no hands-on test is implied.
Source publication: 1997-03-01 · Retrieved 2026-09-16
Establishes that a lens array forms an array of small real images (element images), and demonstrates capturing and displaying them in real time with a television camera, an LCD panel and a lens array, rather than photographic film.
Source publication: 1998-04-10 · Retrieved 2026-09-16
Names the two defects integral photography must correct, pseudoscopic (depth-reversed) images and interference between neighbouring element images, and proposes a gradient-index lens array to address them.
Source publication: 2007-11-01 · Retrieved 2026-09-16
Provides the broader taxonomy (holography, volumetric, multiple-image) that integral imaging's full-parallax multiview approach sits within, without naming integral imaging specifically.
Record: 0009-bf-G1 · Rewrite revision: 2026-09-16.1 · Actual site publication: pending. No exact event date asserted.