Systems‎ > ‎


18 Columns x 4 Rows => 72 customized 46" interlaced passive-stereo digital signage displays arranged in ~6.7m diameter cylinder (June 2012).

    Three Display Models

                  • top-row:          Matrix LX46L-3D 46" custom Xpol shift, top (Planar # PPL-997-6624-00)
                  • middle rows:  Matrix LX46L-3D 46"  commercially available  (Planar # PPL-997-6624-00)      Link      datasheet
                  • bottom-row:   Matrix LX 46L-3D 46" custom Xpol shift, bottom (Planar # PPL-997-6625-00)

A Matrix display unit is modular. Power supply and control circuits are off-display, one module each per four displays. ie. 18 power (2U) and 18 (1U) controller modules for 72 displays. Each display includes a mounting frame and 3-axis alignment system. Interconnection is via CAT6 and 24VDC cables.


Off-Axis Viewing and View Distance

Middle-row displays are identical to the passive-stereo displays installed (Fall 2011) in EVL's 3D Cyber-Commons      profile

CAVE2 top and bottom rows use similar displays customized with shifted registration (-9 degrees and +9 degrees respectively) of the 3D micro-polarizer to accomodate wider off-axis viewing. The vertical stereo viewing angle from the published specs above is thus extended to +/- 26 degrees.


                                  normal polarizer registration                                                                shifted (-/+) polarizer registration for top and bottom rows

The nominal viewing distance in the center of CAVE2 is ~11' (3.3m). At this range shifted displays would not be required as the viewing angle is low. 'Ghosting' or Left/Right channel crosstalk is minimal (< 2%) at this distance, but ramps exponentially beyond a critical vertical angle of view, such as when a user approaches the display surface and gazes high or low. Shifting the micro-polarization layer of the top and bottom displays better aligns the Left-Right pixel views with the viewer, resulting in a wider effective critical angle. This allows much closer Ghost-Free interaction with the display surface than with unshifted displays.

View angle is determined by user eye position (height), view target (pixel location relative to height) and view distance (user distance from the target) within the fixed geometries of the CAVE2 display environment. The Ghost-Free range is approximated for an average viewer in blue below. At closer than about 80 inches, the pixels viewed high and/or low will be outside the critical angle and begin to ghost. On-axis viewing is not affected at any distance.

At view distance < 80" (2.0m) an average height viewer will begin to see ghosting in the lower or upper tiles


The CAVE2 prototype (Feb 2012) was 3 columns x 4 rows, driven by a prototype cluster configuration. The columns were positioned and hinged to represent an arc segment of the varying diameters of full-cylinder layouts under consideration. 

12-tile CAVE2 prototype evaluating shifted display performance in mixed and all-shifted columns

Once +/- 9 degrees was determined as the manufactured shift specification, 12 shifted (6 top shift, 6 bottom shift) display prototypes were ordered. They were combined with unshifted displays to fully evaluate all-shifted (2 top with 2 bottom) columns against mixed-shifted (2 middle-rows unshifted) columns. Beyond the uncertainty of the yield and uniformity of the shifted approach there was also concern with cost (+ 30% shifted), as well as calculated luminance variations expected at the display edges where unshifted and shifted displays would meet. 

                                Ghost level is determined visually by a 'Crosstalk Scale' of interlaced Weissmann test pattern images when viewed in stereo

Visual analysis of the prototype columns determined that the shifted displays did in fact increase the ghost-free range to forecast angles. The expected luminance variations were also observed to be insignificant. Further evaluation determined that cumulative pitch variations in Xpol registration measured from the screen centers to its edges worked in our favor in offsetting detectable luminance variations at the display seams. Independent radiometric measurements made in Planar's optical lab confirmed and quantified those findings. We moved forward with mixed (top and bottom rows shifted, two middle rows unshifted) columns based on these tests. 

In mid-2011 Planar determined to focus their passive-stereo work on their 46" WXGA resolution Samsung LCD panels and rolled-out their 46" 3D Matrix product. They have since introduced 2D HD versions of their Matrix display in both 46" and 55". It remains to be seen if they pursue passive-stereo in the 55" class. It was becoming unlikely that an HD passive-stereo display would become available in the CAVE2 grant timeframe. EVL prepared to move ahead with WXGA resolution in a 46" format.

                          1 x4 evaluation column                                                                        conceptualization of side-by-side evaluation columns

We set-up (May 2011) side-by-side evaluation columns of Planar vs. Hyundai, which had recently entered the market. We purchased the 4 Hyundai displays and also received 4 prototype Planar displays for a short evaluation period. Both products used the same Samsung LCD panel and Xpol material, thus performed very similarly. Initially Hyundai offered a cost advantage, though Planar had advantages with their modular packaging and the fact that they were a domestic partner, aligned to collaborative research and customization, which Hyundai could not match. Over time Planar came to match their price. This was a significant contributor to our selection of Planar displays for our 3D Cyber-Commons in July 2011. Though not a tracked VR installation, the 3D Cyber-Commons served as a worthy CAVE2 pre-prototype.

For more development details, see an excerpt of our 2012 NSF report.


Our stated research goals in 2010 were to produce 1) tiled stereo displays with, 2) ultra-thin bezels at, 3) 55" HD resolution. These were simply unavailable in the marketplace. We evaluated available auto-, active- and passive-stereo technologies. We pursued passive-stereo for several reasons, primarily; maturity/availability of aligned technologies, trends in the 3D market, low cost and complexity of required glasses.

We set-out to build a bench-top passive-stereo display by combining available display products and materials. We purchased:

  • JVC 3D HD display with thick bezels, complete with Xpol micro-polarization
  • JVC 2D HD display Model LT-46P510
  • NEC 2D HD display Model P461
  • Xpol micro-polarization sheet -  46" at HD line resolution

Using a ProScope (w/ polarized lens) to evaluate pixel structure

Our early stereo tests placed the Xpol sheet against the surface of the cannibalized (packaging removed) 2D displays. The displays were on their backs so that we could simply lay the material onto the surface with hand-registration of the polarizer vs. pixel rows. An overhead mirror was used to aid in viewing of the entire display surface at a normal distance.

Observing Left-Right polarization with Xpol material laid onto a 2D display

The commercial JVC 3D display served as the target performance standard. We successfully preserved polarization at varying measured ghost levels though it quickly became evident that a more industrial process was needed to register and bond the Xpol material to the LCD for uniform performance.     Summary Slideset

EVL began working closely with Planar and signed a non-disclosure to pursue development, who in-turn had a relationship with Arisawathe Japanese company that manufactures the micro-polarization (Xpollaminates used by companies such as JVC and Planar to create passive-stereo displays. 

For these and further details see our 2011 NSF Report here as well as a Planar analysis.