The three main techniques for making 3D TV displays that have been commercialized to date are: active glass, passive glass, and auto-stereoscopic (see Figure 1). The first two require special viewing glasses while the third is a glasses-free technology. Each has its own technical strengths and issues. Comparing these different techniques and optimizing their performance requires optical characterization. However, the optical measurement method has to be catered to each technology depending on its specific practical problems.
Figure 1. The three main principles for currently available 3D displays.
Active glass displays alternate between the image for each eye at the highest frequency possible (200Hz or more). The glasses include liquid crystal shutters that are synchronized with the display, allowing each eye to see only its dedicated image. The slow response time of the display and the glasses as well as the imperfect shutter efficiency cause noticeable problems. Because the synchronization between the display and the glasses is optimized for the center, images appear imperfect at the top and the bottom of the displays.
Passive glass displays, meanwhile, show two images (one for each eye) simultaneously but with different polarization states. One image appears on odd-numbered pixel rows and the other on even rows. Viewing glass lenses have circular polarizers that filter out the respective image for the observer. Problems are related to the phase-shift film that is integrated on the display surface to fix different polarization states for the odd and even pixel rows. The two images are never perfectly circular polarized.
Auto-stereoscopic displays, on the other hand, emit two or more images simultaneously along different directions. The observer does not require any additional glasses to experience the 3D effect from the right position. The viewing zone is generally limited in size. In addition, part of the light emitted for one eye falls on the other, inducing ghost effects.
For active glass displays, the idea is to simultaneously evaluate the impact of the gray level transitions, which are a measure of color depth or intensity, and of the temporal synchronization between the display and glasses.1–3 In order to do this, we measured what will be seen by one eye of the observer (we used the right eye for our example) across the glasses with two different images on the display (see Figure 2). The image for the right eye includes five vertical bars with different homogeneous gray levels from black to white. In order to check the impact of the light level on the other eye, the left eye image includes 5×5 vertical bars with the same gray levels.
Figure 2. Images for left eye and right eye with five different gray levels.
We performed measurements with a video-luminance meter, which gives a picture of what the human eye sees across the entire display surface. If the display is perfect, we should see only five homogeneous vertical bars. However, this is not the case because of the ghost effects due to the image on the other eye (see Figure 3). We can clearly determine the impact of the gray levels and of the temporal synchronization. The quantity of parasitic light compared to the normal light (crosstalk) is quantified for 5×5 gray level transitions and along the vertical axis.
Figure 3. One example of measurement on an active glass 3D TV with the corresponding relative gray level variations.
For passive glass displays, we performed two main optical characterizations.4–6 Since the polarization state of the light emitted by such displays depends on the direction, the space usable by observers in front of the display where they can get a good stereoscopic effect must be determined. The second characterization concerns the homogeneity of the emission on the display surface, which is affected by local defects that stem from the process used to integrate the phase-shift film on the glass surface.
The first characterization is made with a dedicated optical system that collects all the light emitted from one spot on the display surface and maps the emission in the angular space. The image shows the white view for the left eye and the black view for the right eye (WK), and the opposite: black view for the left eye and white view for the right (KW). We precisely measured the light that the display emits across the two glass filters versus the incidence and azimuth angles (see Figure 4). The color scale represents luminance. The contrast for left and right can then be computed in a box in front of the display in order to determine the qualified binocular viewing space (QBVS) where the observer must be located to experience the stereoscopy.
Figure 4. Viewing angle measurements on a passive glass display and computed qualified binocular viewing space (QBVS). WK: White view for left eye and black view for right eye. KW: Black view for left eye and white view for right eye.
The above measurement checks the light emitted at a limited number of locations on the display surface, so local imperfections cannot be detected. We performed a second test using the same video-luminance meter used for testing the active glass displays. The system is adjusted along the normal to the center of the display at the recommended working distance. The emission of the entire surface across one glass filter is measured for WK and KW configurations. Frequently, local light leakage can appear on the black image. These defects are related to the imperfect alignment of the phase-shift film with respect to the row of pixels and produce ghost effects for the observer (see Figure 5).
Figure 5. Crosstalk measured on the entire display surface (left) and viewing angle measurements at two particular locations (right).
For auto-stereoscopic displays, we need to be able to predict if the light is going to the left or right eye of the observer. This requires very high angular resolution because the iris of the human eye is relatively small (2–3mm). We have developed an instrument with extremely high angular resolution (better than 0.03°) for this purpose.7–9 With this accuracy, it becomes possible to compute what an observer will see at distances of 5–10 meters, allowing tests of the biggest TVs. We show one example of such a measurement in Figure 6. The computation is similar to what is done for passive glass TVs and provides a QBVS that is generally very small. This also makes possible a comparison of the QBVS characteristics of auto-stereoscopic and passive glass displays.
Figure 6. Viewing angle measurements on an auto-stereoscopic 3D display and computed QBVS.
Going forward, we plan to develop measurement methods for future 3D technologies such as time sequential polarization-based displays and holographic displays.
Pierre M. Boher, Thierry Leroux
Herouville Saint-Clair, France
Pierre Boher joined ELDIM in 2003 and is currently an R&D manager. He has an engineering degree and a PhD in physics. He worked at the French Philips Laboratories and was R&D manager at SOPRA.
Thierry Leroux is the president, CEO, and founder of ELDIM. He has an engineering degree and a PhD in physics. He worked at CEA LETI on the development of different types of displays for ten years before founding his company.
1. P. Boher, T. Leroux, V. Collomb-Patton, Characterization of one time-sequential stereoscopic 3D display -- Part I: temporal analysis, J. Soc. Inf. Disp. 11, no. 2, pp. 57, 2010.
2. P. Boher, T. Leroux, V. Collomb-Patton, Characterization of one time-sequential stereoscopic 3D display - Part II: quick characterization using homogeneity measurements, J. Soc. Inf. Disp. 11, no. 2, pp. 63, 2010.
3. P. Boher, T. Leroux, V. Collomb-Patton, T. Bignon, Optical characterization of shutter glasses stereoscopic 3D displays, Proc. SPIE
7863, pp. 786312, 2011. doi:10.1117/12.870731
4. P. Boher, T. Leroux, T. Bignon, V. Collomb-Patton, Multispectral polarization viewing angle analysis of circular polarized stereoscopic 3D displays, Proc. SPIE
7524, pp. 75240R, 2010. doi:10.1117/12.837509
5. P. Boher, T. Leroux, V. Collomb-Patton, T. Bignon, D. Glinel, Imaging polarization for characterization of polarized based 3D displays, Proc. SPIE
7524, pp. 75241K, 2010. doi:10.1117/12.837529
6. P. Boher, T. Leroux, T. Bignon, V. Collomb-Patton, Multispectral polarization analysis of circular polarizer stereoscopic 3D display, IDW, 2009.
7. P. Boher, T. Bignon, T. Leroux, Auto-stereoscopic 3D display characterization using Fourier optics instrument and computation in 3D observer space, IDW, 2008.
8. P. Boher, T. Leroux, T. Bignon, V. Collomb-Patton, A new method to characterize auto-stereoscopic 3D displays using Fourier optics instrument, Proc. SPIE
7237, pp. 72370Z, 2009. doi:10.1117/12.806774
9. T. Leroux, P. Boher, T. Bignon, D. Glinel, VCMaster3D: a new Fourier optics viewing angle instrument for characterization of auto-stereoscopic 3D displays, SID 11.2, pp. 115, 2009.