A desktop light-field display trades resolution for angle

↗

What the record documents

Looking Glass Factory's own company history page, headed 'Hello, we're Looking Glass', states the company was founded in 2014 and that in 2018 it released what it calls 'the world's first desktop holographic dev kit', the original Looking Glass. That page is current documentation rather than a dated announcement and gives no specific launch day; the company's original 2018 crowdfunding campaign page could not be opened for this entry, so what follows rests on the company's present-day technical documentation rather than on the campaign copy itself. That distinction matters in a field where marketing language and a specification sheet do not always describe the same figure the same way.

What creates the image

The display is a lenticular light-field panel: a sheet of fine cylindrical lenses sits over an LCD so each lens sends light from a strip of pixels beneath it in a slightly different direction. Render enough different strips and a viewer moving side to side sees a continuously changing perspective rather than one flat photograph. The company's developer documentation, in a page titled 'How does Looking Glass Work?', states the display 'provides up to 100 discrete views of a 3D scene' across 'a view cone roughly 58° wide'. Inside that cone, each eye samples a different rendered view and perceives stereo depth with genuine, if limited, head-motion parallax; the documentation does not state what a viewer standing outside the cone sees, though a lenticular design of this kind typically keeps showing its outermost views rather than going blank.

What the quilt format actually is

Content is delivered as a 'quilt': one image or video tiling every rendered view into a grid, read left to right and bottom to top so that, per the documentation page 'Quilts', 'the bottom-left tile...is the leftmost view of the scene, and the top-right tile is the rightmost'. The Looking Glass Portrait model arranges its quilt as an 8-by-6 grid of 48 tiles, fewer than the 'up to 100' views quoted for the wider display family, showing that the exact view count is model-specific rather than one fixed number across the product line. The display's own processing slices the grid apart again and routes each tile to its matching lenticular strip.

Questions to bring to a demo

• How many of the quoted 'up to 100' views does the specific unit on demonstration actually render?
• How is a quilt generated for live or interactive content rather than pre-rendered scenes, and at what computing cost per extra view?
• What happens to image quality for a second or third viewer sharing the cone at once?

The mechanism here is a genuine light-field display in the narrow sense: many real rendered views, not two views duplicated. The precise view count and cone width, though, are figures the company states about its product family in general, not measurements independently taken from a unit for this entry.