Tomas Sluka from C-REAL gave a very compelling presentation at the recent “Seeing is Believing” seminar on the emerging capabilities of their light field AR display. While he described how the display can work as a light field device, he also described a second mode of operation where it can act as a wide FOV 2D display created by tiling 9 separate images.
The light field operation is designed to solve the vergence-accommodation problem by delivering a series of views with slightly different vertical and horizontal perspectives at different depth planes. This creates a volume of pixels at discrete locations that are visually blended by the eye to create very natural looking images.
Physically, the system uses RGB lasers and a ferroelectric LCOS image modulator. This device can run at an 8 kHz field rate. Full color images are created as a series of bit planes with red, green and blue flashed in time sequence. A MEMS scanner is used to change the angle of the images down the optical path. These images are projected onto a holographic optical element (HOE) placed on an ordinary eyepiece. This is believed to be a diffuser HOE and operates in a way that is similar to projector-HOE systems being developed for automobiles. At the event, C-REAL showed the latest prototype that is about 1cc, offers 70-degree FOV with 40 PPD and 2000 nits of luminance at 200 mW.
C-REAL’s Light Field Display Runs at 8 kHz Field Rate to deliver multiple perspectives at various depths
Light field displays can be computationally intensive, but the computations for a near-to-eye light field display are much more efficient than those for a direct-view type light field display as the viewing frustrum is much more limited. As a result, the difference between views can be quite small meaning you don’t need to render each view from scratch, using the previous view to infer pixel changes for the next view. C-REAL also can spread the color rendering bit depth over the views allowing the eye to integrate this into a full-color image.
The newly announced mode, called sequential pixel replication, can be accomplished with a software change. In this mode, instead of creating a light field image, the system can be used to create a 3K resolution 2D image by tiling 9 sub-images into one seamless image. This is done by driving the F-LCOS with a series of image bit planes 2-bit color images and stepping the MEMS multiplexer to the nine positions (3×3 image matrix). When integrated with an eye tracker, a foveated image is determined and more image bit planes are allocated to this region to increase the color depth. This means less color fidelity on the periphery, but that is ok. In this way, the system can run at around a 100 Hz frame rate while creating a 70-degree FOV with a perceived resolution of > 400 PPD. It is a clever application that they have now added to their roadmap as steppingstone to the Holy Grail of true light field displays. Look to see this mode demoed at Photonics West.
C-REAL AR Glasses Roadmap
