Authors: Zhihui Li*, Honglei Ji*,Tianying Deng*
* TCL Electronics Holdings Limited., Shenzhen, Guangdong, China
As all known that TFT-LCD displays has a bad performance of viewing angle, specially compared with OLED displays，LED displays etc..When viewing angle become larger from perpendicular of screen, color become shift and washout. Although thousands of TFT-LCD engineers attempt to improve picture quality, especially viewing angle, by optimizing pixel structure, modifying cell parameter etc.[1,2],the drawback of viewing angle still can’t be overcame.
And it’s also all known that human’s eyes have different perception for different colors, which was well known as MacAdam ellipse[3],but it’s unclear what color is more visually sensitive and easily perceptive for human’s eyes while viewing angle become lager on TFT-LCD displays. Some organizations have different criterions, e.g. China Video Industry Association (the abbreviation is CVIA) recommend nine colors (including human skin color, mixed Red, mixed Green, mixed Blue…) chosen from Macbeth chart to estimate color viewing angle[4], the website of Rtings which focus on product evaluation in North America use pure Red, pure Green, pure Blue colors to estimate color viewing angle etc..So it’s necessary to research what kind of color is more visually sensitive for human’s eyes and if it’s appropriate to estimate viewing angle with pure color or not.
We conducted a perceptive experiment which consists of two parts. The first part was designed to pick colors that have the biggest difference as viewing angle change from perpendicular to oblique out among stacked colors, figure 1 shows human-skin colors by different R/G/B mixing ratios, both with light human-skin colors and dark human-skin colors, figure 2 shows other mixed colors and pure R/G/B color except human-skin color. Participants were asked to select 10 colors which have the biggest difference and try to sort them as possible in figure 1 and figure 2 respectively. In addition, participants should answer a question that if they perceive pure Red/pure Green/pure Blue have some changes in hue and saturation as viewing angle changed, and if the degree of change is within acceptable limits.
The second part was designed to evaluate actual experience in household when LCD display the scene of landscape, human contains various colors, which was shown in figure 3.Participants were asked to judge the perception level as the viewing angle changed from 30 degrees to 60 degrees,15 degrees per step. The perception level was divided into 3 levels named A/B/C.A level means hue and saturation has no or very little change, meaning good grade, B level means hue and saturation has some perceptive change, but the change is acceptable, meaning okay grade, C level means hue and saturation has large perceptive change and the change is unacceptable, meaning poor grade. How to define the A/B/C grade was completely decided by participants. The evaluation results are filled in the form as shown in figure 4.
We used a 55inch TFT-LCD TV with ultra HD resolution(3840×2160) to conduct this experiment, keep the constant distance between participant and TV about 2m,which means a suitable and comfortable distance for the viewer. As said before, viewing angle of participant changed from 30 degrees to 60 degrees, 15degrees per step, 0 degrees as a reference.
A total of 16 participants participated in the experiment, whose own work is backlight design, image quality processing and display technology development etc.
There are three obvious peaks in figure 6,in other words, human-skin colors 1,10,6 are found to have more difference as viewing angle changes by participants, which are close to the color of the race in the figure 8 below, we can name them as dark human-skin colors to distinguish from 3th human-skin color(in figure 1) with a name of deep human-skin color in this paper. So we think maybe human’s eyes are more sensitive to dark human-skin color than light human-skin color and deep human-skin color.
Unfortunately there is no obvious regular distribution in figure 7.we hope to get some regular distribution by changing the way we analyze, that’s the times each color was selected by participants, what we name it as BST(be selected times/total participants).The color has more difference means BST number is more closed to 1,but we also failed to get any obvious regular distribution in this analysis way. So we think human’s eyes are more difficult to percept difference for mixed colors than human-skin colors.
It is important to note that no.29, no.30, no.31 are pure red, pure green, pure blue colors in figure 7 and figure 9.There are 68.8%(11/16) excepted the difference of pure Red/pure Green/pure Blue, 18.8%(3/16) of participants considered the color change to be unacceptable and 31.3%(5/16) of participants didn’t percept pure color difference (as shown in Figure 10). Analysis with Figures 9 and 10, we think using pure red/pure green/pure blue colors to evaluate viewing angle of a TFT-LCD displays is not a suitable way.
We also hope to get some useful conclusions from the analysis of test data. Figure 11 and figure 12 show test data of every color consist of human-skin color, mixed color and pure color. What is interesting in the figure 11 below is that the color (human-skin 3) with greater data changes isn’t more noticeable to the human’s eyes. The result also confirms what we said earlier, human’s eyes are more sensitive to dark human-skin color than light human-skin color and deep human-skin color.
In figure 12,pure red/pure green/pure blue colors have little △u’v’ number as viewing angle changed than mixed colors, This also confirms what we said earlier, using pure red/pure green/pure blue colors to evaluate viewing angle of a TFT-LCD displays is not suitable.
At the second part of perception experiment, the perception level was divided into 3 levels named A/B/C which score 10/8/0 respectively; the final score is the sum of all the single scene score. If final scores are greater than or be equal to 36 and the quantity of grade C is smaller than or be equal to four, the picture quality at this viewing angle is acceptable. Percentage of participants who considered the image quality acceptable is shown in figure 13, all the participants consider picture quality of 30 degrees viewing angle to be acceptable, picture quality of more than 45 degrees viewing angle to be unacceptable. Even when the test result of the viewing angle of 30 degrees is not so good as figure 13, all participants still think the image quality is good, so we think human’s eyes are more inclusive than the test data shows.
We conduct a perceptive experiment to find out human’s eyes are visually sensitive to what kind of color, we found human’s eyes are more sensitive and could accurately perceive color difference to dark human-skin color as viewing angle changes than light human-skin color and deep human-skin color on LCD display, we also found using pure color(pure red/pure green/pure blue) to evaluate viewing angle is unsuitable, because they have small change of hue and saturation and human’s eye can’t easily percept the small difference. At last, we found human’s eyes are more inclusive than the test data shows in fact, test data at this stage cannot represent actual picture quality.
References
[1] Gu-Jin L.,Rui-Jie X. ,I-An Y. ,Chiu-Lien Y. ,JyShan H. ,Evaluating and improving color washout of vertical aligned liquid crystal display, Displays p.159 (2014)
[2] Kun-Cheng T., Chien-Huang L., Shu-En L., Jia-Long W., Feng-Yueh T., Novel Pixel Design to Improve Color Shift of Multi-Domain Polymer Sustained Alignment LCD, SID 2019 Digest, pp 375-378.
[3] David L. MacAdam, Visual Sensitivities to Color Differences in Daylight,Journal of the Optical Society of America, 32, 247-274(1942).
[4] CVIA, Display Performance Methods of Measurement for Flat Panel Television, SJ/T 11348-2016,pp. 6-7.
Click to view the original article:
Symp Digest of Tech Papers – 2021 – Li – 21 3 Research of visually sensitive colors as viewing angle change on TFT‐LCD