A phone can render a three-dimensional game on an ordinary flat screen. A stereoscopic phone does something additional: it delivers different views to the left and right eyes. Glasses-free, or autostereoscopic, displays use optics in the display to direct those views, making viewing position part of the system.
Direct each view toward the intended eye
Because the eyes occupy different positions, they receive slightly different views of a scene. A stereoscopic display supplies an appropriate image to each eye. Sharp's technical explanation describes a parallax barrier: a patterned optical layer controls which display light reaches each eye.
Left viewing zone
Left eye receives left view
Right viewing zone
Right eye receives right view
Microlens designs use another approach to directing light. They should not all be described as parallax barriers. Their common purpose is separating views; their optical construction, efficiency, and switching behavior differ.
The useful viewing regions are often called viewing zones. Moving away from the intended position can mix or swap the views. Crosstalk is unwanted leakage of one view into the other, often perceived as a ghost image. Sharp's engineering paper on switchable 2D/3D displays (PDF) discusses viewing windows and the optical arrangements that create them. A display's dimensions alone cannot describe the resulting viewing experience.
Ask how the two views were made
| Content | What supplies the views? | What to check |
|---|---|---|
| Ordinary 3D graphics | A model is rendered into an image for the screen. | Is there one output view or a separate view for each eye? |
| Captured stereo | Two viewpoints record the subject. | Are the pair synchronized and presented to the correct eyes? |
| Converted 2D | Software derives an additional view from one source image or sequence. | How was depth estimated, and where do edges or hidden surfaces look wrong? |
Conversion cannot be treated as a second camera measurement. Imagine a fence in front of a tree: another viewpoint would reveal different pieces of the tree through the gaps. A conversion system must generate a view despite information missing from the original image. An appealing depth effect and an accurate reconstruction are different outcomes.
What DDD and Ocuity demonstrated in 2006
DDD's May 8, 2006 company announcement described a prototype based on HTC's Faraday handset, using a Texas Instruments OMAP850 processor and Windows Mobile 5. Ocuity added its Polarisation Activated Microlens technology to make the display switchable between 2D and 3D. DDD supplied content software, including real-time conversion of ordinary camera photos and video.
The developers' 2007 engineering paper, by Julien Flack, Jonathan Harrold, and Graham Woodgate, explains the integration of display optics and processing. It documents a prototype demonstration, not evidence that this exact modified Faraday became a mass-market retail phone. Its reported results belong to that implementation and should not be generalized to every glasses-free display.
Make a demonstration repeatable
- Record the device, display mode, screen orientation, and source material.
- Identify whether the content is captured stereo, separately rendered views, or converted 2D.
- Begin at the documented viewing position, then observe what happens with small changes in position.
- Compare 2D and 3D modes using the same material where the device permits it.
- Describe visible depth, ghosting, brightness, and viewing-position limits separately.
A conventional screenshot can document the source pixels or menu settings, but it does not reproduce the optical separation delivered to two eyes. Record that limitation when sharing a demonstration. For another case where display design changed phone use, see the rotating-screen DMB phone.