A 2005 survey split true 3D displays into two working classes

↗

One survey, two named systems

In August 2005, Gregg Favalora published Volumetric 3D Displays and Application Infrastructure in IEEE Computer, reviewing displays that fill an actual volume of space with light rather than presenting a flat or angled image. Its section headings, visible even behind the publisher's paywall, name two working systems side by side: Perspecta, a rotating-screen display, and Depthcube, a stack of fixed panels. Those two names map onto the field's basic split: swept-volume displays move a single surface fast enough that persistence of vision fills a volume, and static-volume displays stack multiple fixed surfaces and switch which one is active.

How each class actually forms an image

The swept-volume approach is documented in detail by the Perspecta patent, assigned to Favalora's own company, Actuality Systems: a semicircular screen spins inside a transparent dome while a stationary projector below updates the image on it 'at least 4,000 times per second'; the patent states plainly that 'persistence of the human visual system integrates these two-dimensional image slices into a three-dimensional volume-filling image'. No glasses or head tracking are needed, and the volume can be viewed from any angle around the dome. The static-volume alternative is documented by the Depthcube patent, assigned to LightSpace Technologies: a stack of liquid-crystal panels switching between transparent and light-scattering states in sequence, each one catching a projected slice at high speed, again fused by persistence of vision.

The shared limit neither design escapes

Both mechanisms genuinely fill a volume with light, which is why they support natural, glasses-free parallax for many viewers at once. Neither, however, produces true self-occlusion: a solid voxel does not automatically block light from a voxel behind it the way an opaque object blocks light in the real world, because each slice is drawn independently. The Depthcube patent addresses this directly, stating the design 'may also avoid occlusion' and describing a workaround of interspersing extra scattering panels to absorb background light. This is a limitation the patent itself acknowledges, not an editorial guess; it means both display classes render translucent-looking scenes unless specifically compensated for, and that compensation is only partial.

Questions to bring to a demo

• Is the volume being swept by a moving surface or built from a fixed stack of surfaces?
• Does a solid object convincingly block the view of what is behind it, or does everything look faintly translucent?
• What is the actual voxel or slice count, and how does resolution fall off away from the display's centre?

Both named systems in the 2005 survey solved 'a volume you can walk around' without solving 'objects that hide each other'. That distinction is worth asking about directly, since a demo lit to hide translucency will not volunteer it.