Analysis of Visual Perception of Cinemagraph Images: Focusing on Q Methodology

Cinemagraphs begin with the approach of the film(cinema), as can be seen from the etymology. It can be seen that the film is a photograph until the movie goes back. The pre-photography era would be painting. As such, the development of images can be understood no matter what fusion content is generated based on historical facts until the era of video today.

The first film began in 1895 with the “cinematographer” filmed invented by the Lumiere brothers on a moving screen. Cinematograph is a combination of the Greek words "Kinema" and "Grafhei." It was originally registered as cynematograghe, but modified ‘Y’ to ‘I’. The beginning of the film began that way, and today the film is easily called cinema. Today, cinemagraphs are being used as fusion contents due to the partial combination of pictures and images, and are newly illuminated in the video era and used in advertising digital devices.

The subjectivity study of this paper is designed and tested by Q methodology. Q methodology was developed by psychologist William Stephenson (1902~1989) as a way of diagnosing a person's mind [1]. This is a way of expressing what subjective research is considered subjectively or by view of card sourcing, factor analysis and estimating the result of groupings to help the understanding of subjective experience. Inmates usually deal with topics such as love, happiness, politics, psychology, religion, and reason, and use them as a way of extracting objectification from the standpoint of subjective views [2].

Therefore, it is a methodology used when using the same clustering of subjects that are difficult to object with social sciences. After gathering the ideas and opinions of the groups and extracting the factors to determine the highest and lowest levels, the views can be specifically defined as statistical values. In general, the Q methodology is an alternative research method that responds to factor analysis, and is used in some parts of the social sciences including communication studies and psychotherapy [3]. It is a typology that objectively classifies people's subjectivity through a unique method called Q-sorting. Therefore, this methodology is used to quantitatively analyze the subjective interpretation felt by watching a cinemagraph.

Another such precedent study is “The Effects of Increased Evocativeness from Visual Stimuli in Online Shopping Mall [4]”, “Analysis of Appreciator's Preference Depending on the Difference in Existence & Unexistence of Movement of Cinemagraph Advertisement” is a study on the preference of visual perception stimulation [5]. “Cinemagraph: the subjective recognition of unfamiliarity” interpreted subjective interpretation in Q methodology and focused on the strangeness of visual perception [6].

There are two types of sample selection—unstructured and structural methods—which are randomly divided. In order to select Q-samples, 32 samples were selected after collecting the discretion of cinemagraphs through expert pre-interviews for 10-year experienced workers, one for each master's and Ph.D. Google randomly searched and collected about 112 cinemagraph images evenly from 'Flixel', a representative application for producing cinemagraphs, for about 1 month from October 1 to October 28, 2019. In this study, Q samples corresponding to 2 * 2 * 2 * 2 * 2 = 32 final statements, such as intensity, number, area, speed, and perceived position of motion, were selected. Through this, negative, neutral and positive opinions were balanced.

This is best evaluated by five groups. The five groups are as follows. ① special interest in the subject; ② dispassionate interest; ③ authoritative or expert in the subject; ④ generally interested in class interest), ⑤ uninterested [7].

The Q methodology focuses on the value of the meaning of subjectivity and deals with differences in importance within individuals rather than differences between individuals. Therefore, the number of P samples is not limited, but it provides an opportunity to express subjectivity to P samplers. Because the purpose of the study is not to produce results for the population, the sampling of the survey subjects is done in a deliberate manner. In this study, P-samples were selected by considering gender, age, education, and occupation based on the criteria presented above. Korean males and females in their 20s and 40s were selected as P-sample subjects with high frequency of using new media and digital images. Then, 39 subjects were interviewed, including 16 males and 23 females, taking into account the Cinemagraph P sample criteria. These test subjects were randomly selected.

For about a month, 39 subjects were interviewed at Sangsu-dong, Mapo-gu, Seoul, Ilsan, Goyang-si, Paju-si, and Yangjeong-dong, Jin-gu, Busan. After setting up the survey schedule, they met with the subjects one-to-one and performed Q classification. The process was to code the prepared digital image directly on the desk, show the image directly on the laptop on the html screen, and then hand the image to the subject. The preferences for 32 cinemagraphs in the Q-Class were divided into three groups: the affirmative group, the neutral group, and the anti-group. Subjects were asked to write the subjective reasons for each of their preferences and preferences. Normally, the Q sample was classified into three groups, which belonged to either the pros, the neutrals, those who were against. After the Q classification, the Q-segment was recovered and coded into a TXT file. The score was calculated by assigning 1 point to the least preferred and 9 points to the most preferred. In other words, when the most negative, (-4) starts with 1 point, 2 points (-3), 3 points (-2), 4 points (-1), 5 points (0), 6 points (1), 7 points (2), 8 points (3) and the most positive were 9 points (4). Fig.1 - Fill in the Q-sorting fields in step 9. Submitted by checking the never ~ 50% ~ Always route. Principal component analysis was performed using the QUANL PC program by reflecting the score results in Q sample number order. Used in Microsoft Windows 7, 32bit PC operating system version.

Fig. 1. Q-Sorting Distribution.

Download Original Figure

On the one hand, the QUANL computer package program used in this study should be theoretically rotated using the Barrimax method. The eigenvalue that determined the number of types was at least 1.00. In addition, correlation coefficients between factors and factor weights were checked to influence the analysis results. After the experiment, the program results were extracted into four types. The eigenvalues of each type were 7.2346, 3.3282, 2.3434, and 1.9315, respectively. The highest number was found in Type 1 because 18 P samples (N = 18) were concentrated in Type 1. And last (fourth type) eigen value exceeding 1.9 seems to be a desirable phenomenon. The explanatory variances by type were 0.1855, 0.0853, 0.0601, 0.495 and the total was 0.3805. That is, the four types have about 38% explanatory power. Since the eigen value or variance of the Q methodology is affected by the number of subjects, it may be relatively insignificant, unlike the R method.

On the other hand, the correlation coefficient between types shows similarity between types. In this study, the correlation coefficient between type 4 and type 2 was 0.432, showing the highest correlation. On the other hand, the correlation coefficient between type 3 and type 2 was 0.220, which was relatively low.

The P-sample between types is shown in the table. Type 1 had several students in their 20s, and they were interested in sample 22. Type 2 is interested in sample 16 and is of interest to office workers. Type 3 is interested in No. 26 and there are many women. Type 4 was interested in sample 40, and there were many men.

The first type is the preferred type for clear movements perceived by the eye. The attention to visual perception is clearly reveals that movement attracts the audience first, who initially focuses on one thing, and then the other which meant that visual attention that is less visible. Unclear movements tend to be unfavorable. Therefore, the movements are seen first and do not care about the rest. In short, it is a type that focuses on clear movements. Preferred images and standard scores were 4 (2.27), 1 (1.98), 10 (1.16), 22 (1.12) and 26 (1.09). On the other hand, unfavorable images and standard scores were 31 (-2.07), 3 (-1.99), and 7 (-1.29), starting from the largest order. In the overall analysis, the image is interpreted as a type concentrating on clear movement through clear and moving boundary lines.

Table 6. Type1-Array of Z-score (±1.00 More Than).

Download Original Figure

The second type is the small motion type, which is about 10% of the movement. I prefer to be perceived as a small movement represented by the cinema graph. This is a surprise after discovering that there was no movement; yet, it was discovered.

The stimulus of movement was changed into a sophisticated expression of the cinemagraph, showing a positive attitude. Preferred images and standard scores were 1 (1.98), 9 (1.33), 5 (1.28) and 26 (1.08). The unfavorable images and standard scores were 3 (-1.98), 31 (-1.81), 30 (-1.48), and 12 (-1.25). Sensory movement is the type of movement in harmony with attitudes stimulated by emotional satisfaction in the overall interpretation.

In the third type, the motion and image correspond. The image was recognized as a video from the beginning which suggests that this type prefers natural rendering.

Table 7. Type2-Array of Z-score (±1.00 More Than).

Download Original Figure

He did not like the unfamiliar image of the strangeness chosen in the non-preferred way and showed a natural attitude with a positive attitude. Preferred images and standard scores were ranked 22nd (1.41), 5th (1.39) and 27th (1.07). By contrast, the non-preferred images and standard scores were 16 (-1.91), 8 (-1.63), 2 (-1.45), 19 (-1.32), 4 (-1.24), and 25 (-1.02). In interpreting the image, it is interpreted as a type that prefers common sense, scientific rationality, and familiarity, and prefers an unstable state.

They prefer strong, fast images, prefer artificial, colorful movements, and prefer dynamic movements. The image is visually strong, prefers movement with wide speed and an active area, and prefers artificial movement. The preference images and standard scores were 29 (2.50), 21 (1.83), 24 (1.46), and 15 (1.13). On the contrary, the unfavorable images and standard scores were 3 (-2.00), 31 (-1.69), 27 (-1.38), and 30 (-1.19). When viewing an image, it is interpreted as a type that prefers colorful and strong contrast effect and fast and strong movement.

Table 8. Type3-Array of Z-score (±1.00 More Than).

Download Original Figure

The types that were excluded from the experiment and not extracted must also be outlined. In contrast to the third type, it is a type that shows an attitude of preferring an unfamiliar movement. Regarding 19 or 8, ID No. 1 said, “The narrative element of movement was good. The man to be moved was stopped and screaming only as a girl stimulates the imagination and makes me feel good.” Regarding number 5, subject ID 17 explains the unextracted type, “I expressed a fresh point of view that I had never imagined this place moving”.