2.1.

Pixel Design

Efficiency for light utilization (transmittance) in LCDs is the most significant feature, which is the strength of LCDs compared with OLED displays. Brightness and power consumption of LCDs are strongly influenced by transmittance. However, improving the performance of LCDs inevitably causes a decrease in transmittance, and there is a trade-off in improving image quality. That is, a decrease in the aperture ratio when the resolution is higher, or a decrease in transmittance of color filter when the color is deeper.

The IPS technology at Panasonic liquid-crystal display (PLD) has evolved on the basis of improvements in transmittance, as shown in Fig. 3. In the beginning when PLD started the development of LCDs, the transmittance of an IPS mode was even lower than that of other LC modes with a view to only a wide viewing angle that was a characteristic of IPS. In 2002, the transmittance of AS-IPS pixel, which had the transparent electrodes for both pixel and common, increased by 30%. After that, we developed the IPS-Pro structure, which improved the contrast ratio.^11–13 The pixel of IPS-Pro has a finger-on-plane, which is composed of transparent finger electrodes on a transparent common electrode plate. The IPS-Pro structure was applied to liquid-crystal panels for TVs for the first time in the world. The transmittance was twice as high as at the start of the development, which realized to take the expensive film that enhances brightness out of the LCD. At last, the IPS-pro-next structure was developed, which improves transmittance dramatically owing to spread the aperture ratio of the pixel.¹⁴ As a result, the transmittance has been enhanced three times as the initial introduction of the IPS technology. Especially, as IPS-pro-next has the advantage of transmittance compared with other liquid-crystal modes when the resolution of LCDs is higher, it became the de-facto standard for display devices exceeding 200 ppi such as tablets and smartphones.

Fig. 3

Historical development of IPS technology in PLD.

The pixel size of an 8K4K panel with the diagonal size of 55 in. is 160 ppi, which is that of the same as a tablet of several years ago. Consequently, it is appropriate to apply IPS-pro-next with shield common electrode to the pixel of the 8K4K panel as shown in Fig. 4. This innovative pixel structure has developed by an evolution from IPS-pro that has production results. The Himeji factory has enough technique to produce high-quality IPS-pro panels. Therefore, we were able to achieve large size panels of 55 in. with the IPS-pro-next structure that had been applied to small high-definition panels.

Fig. 4

Comparison of pixel structures (a) without and (b) with a shield common electrode.

2.2.

Driving Technology

2.2.1.

Issue to drive an 8K4K panel

The key issue of note is that the charge time in a pixel is short because of the 8K4K-LCD-driven 120 Hz. The electric mobility of a-Si TFTs is extremely low, around 0.3 to 0.5 (cm/Vs), which the pixel is charged insufficiently. The most straightforward way to solve this charge shortage is the multiplication of gate-scan so as to double the charge time.

• (a) It separates the screen vertically, and drives each screen simultaneously as shown in Fig. 5.

• (b) It doubles the number of data-line, and drives two consecutive gate lines simultaneously as shown in Fig. 6. This architecture is called 1G2D (1 gate and 2 data) pixel.

Fig. 5

Schematic panel architecture of the gate-scan method in the liquid-crystal panel: (a) gate single-scan and (b) gate multiscan.

Fig. 6

Relations of the data supply line in case of (a) 1G1D pixel and (b) 1G2D pixel.

1G2D pixel requires twice as much as the data driver required for 1G1D pixel, meaning 96 pieces with a standard data driver that is 960 output pins. Furthermore, with even 1440 output pins, the data-driver requires 64 pieces as well. Thus, it is more difficult to realize a small size panel because 8K4K-LCDs need to mount a lot of drivers. For example, the size of 70 in. or more is required to mount 64 data drivers in a panel as shown in Table 2.

Table 2

The number of COF and the panel size which can be equipped with.

Attributes	Units	Specification
Pixel architecture		1G1D	1G1D	1G2D	1G2D
Channel	Num	960	1440	960	1440
Number	Pcs.	48	32	96	64
Mount limit panel size	inch	45 to 55	30 to 40	80 to 90	70

Moreover, 1G2D pixel has a weak point on an aperture ratio in case of high definition because the nontransparent area of wiring metal increases (Fig. 7). As described in Sec. 2.1, there is a trade-off in improving image quality. The loss of transmittance, which is the most important feature of liquid crystal, is one of the issues of 8K4K-LCDs.

Fig. 7

An aperture ratio loss of the 1G2D structure for the 1G1D structure.

In short, 1G2D pixel has some disadvantages, not only the number of drivers mounted, but also the decrease in the aperture ratio. Therefore, these are critical issues for realizing the compact 55-in. 8K4K-LCD. 8K4K-LCDs must be achieved with 1G1D pixel to solve these issues. Adaptive precharge driving (APD) scheme, which is a specific technology of PLD, enables realizing the 1G1D pixel.¹⁵

2.2.2.

Innovative adaptive precharge driving technology

In this section, we introduce the principle of APD that improves pixel charging performance. The voltage waveform that charged the data to the pixels in the active matrix driving of LCDs is explained with the pattern displaying the white window on the black background as shown in Fig. 8(a).

Fig. 8

APD technology compensating the charge shortage of the pixel: (a) display pattern and the position of the pixel to be considered, (b) voltage waveform of a pixel(A) with single gate, (c) voltage waveform of a pixel(A) with double gate, (d) voltage waveform of a pixel(B) without APD, (e) voltage waveform of a pixel(B) with APD, and (f) voltage waveform of a pixel(B) with advanced-APD.

The state of driving voltage at a pixel(A) is shown in Fig. 8(b), and the voltage charged to the pixel(A) is insufficient because the time of charging to the pixel in 8K4K-LCDs is too short. Therefore, first, as shown in Fig. 8(c), the time of gate-on is doubled (double gate). As a result, the pixel voltage is sufficiently charged. Charging at the pixel(A) is a best-case condition because of a combination of column inversion driving¹⁶ and double gate. The data voltage is not changed during the period of gate-on owing to display the same white at the previous pixel and the current pixel. Hence, the charging of the pixel(A) is done easily.

By contrast, charging at the pixel(B) is a worst-case condition. In pixel(B), display data are changed from black to white unlike pixel(A). Therefore, there is no effect of double gate lines selection, and the pixel(B) is charged insufficiently as shown in Fig. 8(d). Second, we have proposed a solution to this difficult issue in addition to double gate. As shown in Fig. 8(e), gate-on time is divided into two parts, and precharged voltage is applied during the former half of the time. The value of the precharged voltage for a combination of all gray-scale levels has been obtained by pixel voltage simulation using SPICE and automatic measurement by 2-D color analyzer (CA-2500).

We applied this APD technology to all of mass production TVs driven at 240 Hz, and it performed adequately with high reliability. However, in terms of improvements desired in conventional APD, there were the following two issues:

• (1) The voltage required for precharge is high.

• (2) The data driver with the twice of the speed is essential to charge a precharge voltage and a real voltage within a period (1H).

To solve these problems, we propose advanced-APD as shown in Fig. 8(f).^17,18 Inputting an adjusted precharge voltage constantly during 1H period is enabled to reduce a compensation voltage. Moreover, the speed of data does not have to be doubled like conventional APD.

2.2.3.

Simulation result of adaptive precharge driving

The optimum value of the precharge voltage is determined by charge simulation of the pixel voltage. The simulation tool is based on SmartSpice produced by Silvaco, Inc., and as shown in Fig. 8, the voltage charged to the pixel capacity is calculated by inputting the data voltage through a-Si-TFT of which the model parameters are tuned by ourselves. In the case of pixels optimized for 120 Hz, the size of TFT is too large. Therefore, aperture ratio is reduced and the load of the wiring also increases. Hence, we calculated the precharge voltage to realize 120 Hz with TFT of the comparatively small size optimized at 60 Hz. The ratio of the precharge voltage for the current data voltage in conventional APD is shown in Fig. 9(a). In contrast, Advanced-APD shown in Fig. 9(b) is enabled to make the precharge voltage low as expected. In this graph, data7 shows white data, and data7 shows black data. Precharge ratio ($P_{\text{ratio}}$) is defined as follows:

Eq. (1)

$$P_{\text{ratio}}=\frac{V_p-V_d}{V_d},$$

where $V_p$ is the precharge voltage level in the case that the pixel voltage level of Fig. 8(c) is equal to that of Fig. 8(e) or Fig. 8(f), and $V_d$ is the data voltage level in current period of Fig. 8. The precharge ratio was required 26% of white voltage in conventional APD when data changed from black to white. In advanced-APD, we were able to obtain the result that the pixel capacitance was charged enough with precharge ratio of slightly over 13%.

Fig. 9

Precharge ratio (the ratio of the precharge voltage for the current data voltage) comparison between (a) conventional APD and (b) advanced-APD.

2.2.4.

System architecture of adaptive precharge driving

The system architecture of advanced-APD is shown in Fig. 10, which is the same as conventional APD. Main components are line memory and look-up-tables. The optimum precharge voltage value is in proportion to difference voltage of the gray-scale levels, an exponential function of a distance from a data driver and the reciprocal of panel temperature, that is, the pre-charge voltage level is given as follows:

Eq. (2)

$$V_p=V_d+V_{dd}{e}^{-F(i_{\mathrm{TFT}},t_H,C_{\mathrm{pix}})}{e}^{-G(C_d,R_d)},$$

where $V_{dd}$ is the voltage obtained by subtracting the data voltage of the previous line from that of the current line, $i_{\mathrm{TFT}}$ is the TFT-ON current, $t_H$ is the period of 1 H, $C_{\text{pix}}$ is the pixel capacitance, $C_d$ is the data-wire capacitance, and $R_d$ is the data-wire resistance. The function $F$ express TFT charging characteristics, and $i_{\mathrm{TFT}}$ is related to panel temperature. The function $G$ express electrical time constant of data wire, and $R_d$ is related to a distance from a data driver. All of these data are stored in the electrically erasable programmable read-only memory as a large number of tables.

Fig. 10

System architecture of APD.

2.3.

Wide Color Gamut

2.3.1.

Color gamut corresponding to the light source

The benchmark result of color gamut with various light sources toward BT.2020 is shown in Fig. 11 and Table 3.

Fig. 11

Various light sources for LCDs and the peak wavelength (a) B-LED + RG-phosphor, (b) B-LED + QDs, and (c) RGB-all-laser.

Table 3

Benchmark results of color gamut (u′v′ covering) by representative color technology.

Devicetype	Specification
Color target	DCI-P3/Adobe RGB	BT.2020
OLED for TVs	85%	65%
LCD with B-LED + RG-phosphor	100%	82%
LCD with B-LED + QDs	100%	87%
LCD with three primary color laser diodes	100%	99%

LCDs have been able to cover Adobe RGB and DCI-P3 with conventional color technologies. The color gamut of LCDs is wider than that of OLED displays owing to functional parts are separated by a light source of backlight and a color filter. Nevertheless, white LEDs based on blue LEDs, which contain the red and green phosphor, cover only about 82% for BT.2020. Because full width at half maximum (FWHM) of the wavelength of the blue LED is as wide as $\sim 55\mathrm{nm}$ and that of the red and green phosphor are very broad, blue and green, red and green are mixed. Therefore, it is really difficult to widen the color gamut with a combination of the blue LED and the phosphor. FWHM must be $<10\mathrm{nm}$ to fully cover this BT.2020. The wide color gamut technology using quantum dots (QDs) has been the most popular recently. Although the FWHM of QDs with toxic cadmium is as narrow as $\sim 20\mathrm{nm}$, that of QDs without cadmium is $\sim 40\mathrm{nm}$. In case of this FWHM, QDs can express only colors slightly deeper than LEDs with phosphor. Ultimately, now, to cover almost 100% of BT.2020, we have no choice but to apply laser diodes of three primary colors to a backlight. Each FWHM of laser diodes with three primary colors are 2 nm or less, and each color is reliably split.

In this work, we achieved wide color gamut of the 8K4K panel in two stages. First, as a realistic solution, we realized to cover 100% of DCI with a white LED. In the next stage, as a future solution, we realized to cover close to 100% of BT.2020 with laser diodes of three primary colors.

Next, we will explain a laser diode.

2.3.3.

Laser diode backlight

In the case that laser diodes are applied to the backlight of LCDs, another issue is to make surface light sources. Although LED has Lambertian distribution, laser diodes have strong directivity. In the case of LED backlights, the light is directly input to a light-guide-plate as shown in Fig. 12(a). Contrary to this, in the case of laser diode backlights, a simple combination of laser diodes and conventional light-guide-plate cannot avoid Mura as shown in Fig. 12(b).

Fig. 12

Backlight structure with laser diodes as the light source (a) LED backlight, (b) laser-diode backlight, and (c) laser-diode backlight with segmented light-guide-plates added.

We solved this problem by developing the combination of multiple light-guide-plates as shown in Fig. 12(c). Once the light is diffused by the added segmented light-guide-plates, the diffused light is input to a conventional light-guide-plate from panel edge.

As a result, the luminance uniformity of the backlight with laser diodes can be made equivalent to that of the backlight with LEDs.

2.4.

Effect of Prototype Sample

We achieved the 55-in. 8K4K prototype sample in the mass production line of 8.5 generations in Himeji. As shown in Fig. 13(a), color shift of IPS mode is not visually recognized at all at the viewing angle of 100 deg. Fig. 13(b) shows a monitor with a VA mode, which is currently sold in the market for comparison purposes. Thus, it changes to whitish colors at the viewing angle of 100 deg.

Fig. 13

Viewing angle characteristics in case of (a) IPS mode and (b) VA mode: (Left) 50 deg viewing angle, (center) 0 deg viewing angle, and (right) $+50\mathrm{deg}$ viewing angle.

2.4.1.

Advanced-adaptive precharge driving image quality

In such a way as to confirm the effectiveness of APD, we compared front view photographs of character image displayed on a 55-in. 8K4K panel driven 120 Hz as shown in Fig. 14. The displayed image is black characters on a gray background with 50% brightness, which makes it easy to recognize the difference visually. Brightness of the gray pixel of the line next to the black pixel on the LCD driven without APD was decreased. Consequently, the characters looked smeared. The pixel of which brightness became dark is definitely caused by the shortage of charging from the pixel TFT.

Fig. 14

An example of APD effect.

Table 4

Specifications of prototype 55-in. IPS-LCD.

Attributes	Units	Specification
Color target		DCI-P3/Adobe RGB	BT.2020
Display diagonal size	inch	55	55
Native resolution		8K4K $7680\times 4320$	8K4K $7680\times 4320$
LC alignment process		Photoaligned IPS	Photoaligned IPS
Display colors	bit	10	10
Pixel pitch	mm	0.1575 (160 ppi)	0.1575 (160 ppi)
Data driver count	pcs.	48	48
Luminance	nit	400	350
Contrast ratio		1500:1	1500:1
Picture frame rate	Hz	120	120
Color gamut ($u^{\prime }{v}^{\prime }$ covering)		100% (DCI/Adobe RGB) 82% (BT.2020)	100% (DCI/Adobe RGB) 99% (BT.2020)
Power consumption	W	180	350

Fig. 15

Table type 55-in. display in International CES 2015.

Fig. 16

55-in. 8K4K display realized color gamut covering 99% $(u^{\prime }{v}^{\prime })$ of BT.2020 with the laser diode in International CES2016: we collaborated with the connection technology of plastic optical fiber.¹⁹

In contrast, APD was enabled to obtain a clear outline of the character because it fully charged each pixel capacitance in the panel.

It turned out that APD is an innovative technology that has the effect of improving the charging performance of TFT.

3.1.

Effect of Three-Dimensional Technology in Medical Field

Camera and monitoring systems that support surgeon’s vision are one of the most promising advanced medical technologies. These systems are an endoscope and a microscope surgery system, and expanding recently. However, accident such as hurting organs with surgical instruments during operations is concerned, and stereoscopic is required in these procedures, as shown in Fig. 17(a). Furthermore, a medical application of 8K4K technology has been enabled to recognize the details of organs. Medical Imaging Consortium succeeded in the world for the first time in clinical application applying 8K4K imaging technology as shown in Fig. 17(b) in 2014.³ As a result, it was demonstrated that surgeon was able to recognize clearly the details of diseased parts. Moreover, the imaging of 8K4K has wide viewing field, and can prevent the collision of surgical instruments and an endoscope during the operation.

Fig. 17

Cameras and monitors support surgeon’s vision: (a) 3-D heads-up surgery for ophthalmology and (b) 8K endoscopic surgery, [Photograph (a) is provided by Takashi Koto, Kyorin University, the department of ophthalmology].

In this section, we describe a stereoscopic technology based on 8K4K IPS-LCDs.

3.2.

Three-Dimensional Display System with a Polarization Filter

The mechanism of stereoscopic is that a right-eye-image and a left-eye-image are observed by a right eye and a left eye, respectively. In general LCD-TVs, there are two approaches for realizing stereoscopic. One is a frame sequential driving, which is based on a time division of a right-eye-image and a left-eye-image. The other is a 3-D polarization filter on a surface of a LCD, which is based on a space division of those images. It is important for medical applications to reduce fatigue of surgeons. Hence, the space division technique has adopted by surgeons for the reason of some advantages, such as lightweight, flicker-less, and battery-less eyeglasses. However, this technique has two issues as described below.

The vertical resolution in 3D-LCDs based on the space division is half of that in 2D-LCDs in essence. Thus, the detail of image is much spoiled in the case that the resolution of LCD is low. Typical images output on a FHD-3D and an 8K-3D display are compared in Fig. 18. Edges of objects on the FHD-3D display are jagged, for this reason, the detail of structure is lost. This is the first issue, “reduction of vertical resolution.” By contrast, the 8K-3D display is enabled to avoid severe artifacting of edges of objects.

Fig. 18

Comparison of image resolution displayed on the FHD-3D and the 8K-3D.

In the situation that surgeons observe at right in front of a 3-D display, they are able to recognize stereoscopic. However, in the situation that the display is observed from tilted angle in the vertical direction, the left- and right-eye-images are overlaid in eyes of surgeons as shown in Fig. 19. This mixture of each eye’s image is known as 3-D crosstalk. In this case, the separated left- and the right-eye-image reach both eyes. This is the second issue, “restriction of viewing angle along the vertical direction.”

Fig. 19

Limited viewing angle along vertical direction.

The cause of occurrence of 3-D crosstalk is shown in Fig. 20. In the situation of observation at right in front of a 3-D display, the light beam from a left-eye-pixel of an LCD reaches the left-eye of the observer through a filter for the left-eye. The light beam of a right-eye-pixel is also the same. Therefore, the observer can recognize 3-D images correctly. In contrast, in the situation of observation at any tilted angle, a part of the light beam from the left-eye-pixel reaches the right-eye of the observer through a filter for the right-eye. The origin of the 3-D crosstalk phenomenon is the mixture of the light beams.

Fig. 20

3-D crosstalk phenomenon.

Meanwhile, the debasement such as reduced resolution and 3-D crosstalk is not observed from the horizontally tilted angle at all. A wide viewing-angle characteristic, which is a merit of IPS-LCDs, can be obtained completely along the horizontal direction.

3.3.

Optimal Viewing Position

In the stereoscopic based on a space division technique, there are limitations on the position where the whole screen can be observed clearly due to 3-D crosstalk. Furthermore, the advantages of high resolution of 8K4K are not obtained even if the viewing distance is too close or too far away. Thus, the viewing distance from a 3-D monitor need to be decided to design a 3-D filter properly. We decided the distance between a surgeon and an 8K-3D monitor to be 1.0 to 1.5 m, which considered practical examples of 3-D head-up surgery for ophthalmology as shown in Fig. 21(a). In addition, the 3-D monitor is located at 1.0 to 1.5 m apart from the surgeon, such as the layout of the operating room for ophthalmology is shown in Fig. 21(b). This space is required to set an operating table. In general, the defined viewing distance for 8K-2D displays, as described in Sec. 1, is 0.75 times the screen height. In contrast, the optimal viewing distance for 8K-3D displays based on the space division technique is 1.5 times its height due to reducing the vertical resolution by half as shown in Fig. 22.

Fig. 21

Viewing distance of 3-D monitor (a) 3-D heads-up surgery for ophthalmology and (b) example of operating room layout (top view), [photograph (a) is provided by Takashi Koto, Kyorin University, the Department of Ophthalmology].

Fig. 22

Optimal viewing distance of 8K.

We drew a conclusion from the studies discussed above that the diagonal-size of 55 in. was suitable for medical 8K-3D.

We simulated the 3-D crosstalk ratio with ray tracing simulation to fabricate the 3-D polarization filter. The mixture ratio of unsuitable beams included in the output beams was expressed by the 3-D crosstalk ratio. The required dimensional parameters for the simulation are shown in Table 5. In this work, some part of the parameters were fixed, and the pitch of a 3-D polarization filter was changed corresponding to various conditions of the viewing distance of 1.0, 1.2, and 1.5 m. The estimated region of stereoscopic is shown in Fig. 23. We set the maximum 3-D crosstalk ratio to 7%.²⁰ Namely, the position where the 3-D crosstalk ratio is $<7\%$ is expressed by the diamond-shaped areas in Fig. 23. Surgeons can recognize clearly stereoscopic image in these areas. The region of stereoscopic vision is 0.8 to 1.3 m, 0.9 to 1.5 m, and 1.2 to 2.5 m, respectively, for the viewing-distance of 1.0, 1.2, and 1.5 m. As a result, we decided the suitable viewing-distance as 1.2 m covering the range of 1.0 to 1.5 m. In this case, the vertical viewing angle was calculated 8.2 deg, and when stereoscopically viewable area is converted to height, it corresponds to $\sim 17\mathrm{cm}$. In the case of 3-D heads-up surgery, surgeons and assistants are sitting in a chair and can recognize stereoscopic viewing because the height of the viewpoint is stable. As described in Sec. 3.2, there is no constraint on the 3-D viewing angle with respect to the horizontal direction. Therefore, stereoscopic viewing is obtained at the same time not only by surgeons, but also by assistants.

Table 5

Dimensional parameters for the simulation.

Fig. 23

Estimated region of stereoscopic along vertical direction.

We manufactured 8K-3D IPS-LCDs with parameters in Table 5 and the target viewing distance of 1.2 m. The results of measuring 3-D crosstalk are shown in Fig. 24. The vertical viewing angle with the 3-D crosstalk ratio of 7% or less is $\sim 8.6\mathrm{deg}$ (Fig. 24). The measurement and the simulation of 3-D crosstalk are matched accurately. Thus, we confirmed that the image quality of stereoscopic can be estimated before manufacturing.

Fig. 24

Results of 3-D crosstalk.

3.4.

Image Quality

First, we demonstrated the stereoscopic image on the 8K-3D LCD panel to $\sim 50$ people including surgeons, persons from medical-equipment- and video-equipment-suppliers, and interviewed them for the efficacy and impressions about the image quality. We displayed 3-D computer graphics (CG) for demonstrations. The CG, which was constructed 3-D model data such as laser scanned skeleton specimens and 1 million polygons, was rendered at 8K4K resolution. Observers made some comments that “pixels cannot be seen at all,” “it seems to be natural, and the eyes do not get tired,” and “real things are in front of me.” Next, we had several demonstrations at various medical conferences and exhibitions using real 3-D images taken with 8K4K camera as shown in Fig. 25(a). As a result, we obtained a high rating about the image quality.

Fig. 25

The 8K-3D LCD (a) demonstration of 8K-3D, (b) example of 8K-3D images, and (c) a binocular stereoscopic microscope with two 8K4K cameras.

An example of 8K-3D images shown in Fig. 25(b) is the situation when we tried sutures with endoscopic surgical instruments. First, we took this image using two 8K4K cameras simultaneously, which were developed by JVCKENWOOD Corporation and Kairos Co., Ltd., mounted on a binocular stereoscopic microscope (Nikon Corporation, SMZ1270 and Plan Apo 0.5x/WF objective lens) for the first experimental model of an 8K-3D medical microscope in the world as shown in Fig. 25(c), after that converted to the 3-D image format for the space division technique. These 3-D images on the 8K-3D LCD provide the sense of depth and reality for observers. From now on, we are planning to evaluate the clinical application of the 8K-3D system with this IPS-LCD, such as an 8K-3D medical endoscope/microscope.