본 연구에서는 국내에서 성공한 영화들의 쇼트 분석을 통해 흥행한 영화의 클라이맥스부분에서 공통된 편집 패턴 분석을 찾아 씬(작은 이야기 단위)의 구성이 잘 기획되어 있는지 데이터 시각화 연구를 한다. 이 연구는 편집패턴들의 모형을 참조하여 영화 전체에 클라이맥스 표현 패턴이 몇 개로 구성되어 있는지 분석하는 것으로 쇼트이미지들의 자동 수집과 수집된 데이터들의 샷사이즈 자동 분류 시스템을 설계하고 이 시스템을 통해 클라이맥스 패턴 중심으로 하나의 씬을 이루고 있음을 증명한다. 작은 이야기인 씬의 구성이 클라이맥스 패턴으로만 판단하기 어려워 배우들의 대화를 통해 씬을 찾아 비교분석을 하였다. 배우들 간의 대화 기반 씬 예측을 위한 character-net은 등장인물들 간의 대화 내용을 추적하여 인물들 간의 대화 형성을 네트워크 망 모양으로 시각화할 수 있다. 망 모양의 시각화를 통해 큰 이야기와 작은 이야기의 구성을 분석할 수 있으며, 씬 수에 따른 밀집도로 영화의 흥행 여부를 예측할 수 있다. 이 두 가지 연구를 비교하여 영화의 기획 구성 및 제작 방법에 기여를 할 것이라 판단한다.
본 연구에서는 국내에서 성공한 영화들의 쇼트 분석을 통해 흥행한 영화의 클라이맥스부분에서 공통된 편집 패턴 분석을 찾아 씬(작은 이야기 단위)의 구성이 잘 기획되어 있는지 데이터 시각화 연구를 한다. 이 연구는 편집패턴들의 모형을 참조하여 영화 전체에 클라이맥스 표현 패턴이 몇 개로 구성되어 있는지 분석하는 것으로 쇼트이미지들의 자동 수집과 수집된 데이터들의 샷사이즈 자동 분류 시스템을 설계하고 이 시스템을 통해 클라이맥스 패턴 중심으로 하나의 씬을 이루고 있음을 증명한다. 작은 이야기인 씬의 구성이 클라이맥스 패턴으로만 판단하기 어려워 배우들의 대화를 통해 씬을 찾아 비교분석을 하였다. 배우들 간의 대화 기반 씬 예측을 위한 character-net은 등장인물들 간의 대화 내용을 추적하여 인물들 간의 대화 형성을 네트워크 망 모양으로 시각화할 수 있다. 망 모양의 시각화를 통해 큰 이야기와 작은 이야기의 구성을 분석할 수 있으며, 씬 수에 따른 밀집도로 영화의 흥행 여부를 예측할 수 있다. 이 두 가지 연구를 비교하여 영화의 기획 구성 및 제작 방법에 기여를 할 것이라 판단한다.
This study conducts data visualization of common climax patterns of Korean blockbuster films to analyze shots and evaluate scene (subplot unit) arrangement. For this purpose, a model of editing patterns is used to analyze how many climax patterns a film contains. Moreover, a system, which automatica...
This study conducts data visualization of common climax patterns of Korean blockbuster films to analyze shots and evaluate scene (subplot unit) arrangement. For this purpose, a model of editing patterns is used to analyze how many climax patterns a film contains. Moreover, a system, which automatically collects shot images and classifies shot sizes of collected data, is designed to demonstrate that a single scene is composed based on a climax pattern. As a scene is a subplot and thus its arrangement cannot fully be analyzed only by climax patterns, dialogues of starring actors are also used to identify scenes, and the result is compared with data visualization results. It detects dialogues between particular actors and visualizes dialogue formation in a network form. Such network visualization enables the arrangement of main subplots to be analyzed, and the box office performance of a film can be explained by the density of subplots. The study of two types comparison analysis is expected to contribute to planning, plotting, and producing films.
This study conducts data visualization of common climax patterns of Korean blockbuster films to analyze shots and evaluate scene (subplot unit) arrangement. For this purpose, a model of editing patterns is used to analyze how many climax patterns a film contains. Moreover, a system, which automatically collects shot images and classifies shot sizes of collected data, is designed to demonstrate that a single scene is composed based on a climax pattern. As a scene is a subplot and thus its arrangement cannot fully be analyzed only by climax patterns, dialogues of starring actors are also used to identify scenes, and the result is compared with data visualization results. It detects dialogues between particular actors and visualizes dialogue formation in a network form. Such network visualization enables the arrangement of main subplots to be analyzed, and the box office performance of a film can be explained by the density of subplots. The study of two types comparison analysis is expected to contribute to planning, plotting, and producing films.
* AI 자동 식별 결과로 적합하지 않은 문장이 있을 수 있으니, 이용에 유의하시기 바랍니다.
문제 정의
The character-net creates a network of characters(actors), accumulates their dialogues, and then identifies an actor-based storyline by using the share of dialogue concerning a particular event between actors [6]. In order to find out the secrets of well-made scenarios and the production of blockbusters, this study focuses on how many stories are developed in the network of actors. Figure 2 is a statistical representation of actors' relationships shown in a sequence from Dancing with Wolves, which was obtained by using the character-net.
제안 방법
Consequently, this study performs a comparative analysis between the number of climax patterns, which are automatically classified, and the network diagram based on actors' dialogue, which is obtained by using the character-net.
Millions of shot image data sets are needed to analyze how many scenes containing a climax pattern exist in a single film. For this reason, this study sets up a deep learning model that can automatically analyze shot sizes of image data. Among many types of deep learning like GAN and RL , Convolutional Neural Network (CNN) is used in this study, as it is most widely used in image processing.
Accordingly, a sequence is defined as a main plot gathered many subplots. However, as the arrangement of scenes, which are subplots, cannot be fully identified by climax patterns, this study also utilizes a character-net for comparative analysis in order to show that a subplot is organized around a climax and to predict that a single subplot constitutes a scene. The character-net detects dialogue between particular actors and visualizes dialogue formation in a network form.
In the field of intelligent systems, the latest trend is collecting and analyzing big data to find out patterns in human-made content and applying the patterns thus discovered to machine learning for automatic processing of complex and difficult contents. This study also utilizes the continuity editing rule, which is one of editing techniques, to analyze film content. Many films and TV dramas commonly show climax patterns that break the continuity editing rule.
This study analyzes Dancing with Wolves, Ode to My Father, and Thieves by visualizing subplot arrangements from two perspectives. Satisfactory accuracy has not been achieved in analyzing the number of subplots.
As sufficient shot image data could not be gathered manually, an automatic system was designed that collects shot images and classifies shot sizes [3]. This study conducts a visualization analysis to detect how many climax patterns are extracted in a single film and demonstrates that a single scene is organized based on a climax pattern. In films and TV dramas, a scene is a subplot, and multiple subplots form a sequence.
이론/모형
In this study, shots of some Korean blockbusters are analyzed and common editing patterns are identified by the data visualization method. In the beginning, shot data were manually collected by using Daum PotPlayer, and the camera shot sizes were classified to derive regular patterns [2].
후속연구
This success is attributable to well-organized integration and implementation of scenario planning, shooting, and editing skills, and special effects. To contribute to the continuous development of Korean Wave content, this study attempts to visualize scenarios and editing techniques for big data analysis. Our attempt is a preliminary work of building up an intelligent film analysis system that will support preliminary analysis of scenarios, scripts, and continuities before producing films or TV dramas.
참고문헌 (11)
I. Arev, H. S. Park, Y. Sheikh, J. Hodgins and A. Shamir, "Automatic editing of footage from multiple social cameras," in Proceedings of ACM SIGGRAPH- ACM Transactions on Graphics (TOG), Vol. 33 Issue 4, July 2014
Y. M. Lim, "The Climax Expression Analysis Based on the Shot-list Data of Movies," The Journal of Broadcast Engineering, Vol.21, No.6, pp. 965-976, November 2016.
J.E. Eom, S. R. Park, Y. M, Lim, "Design of Deep Learning for Movie Shot Size Recognition," in Proceeding of the Korea Multimedia Society, Busan, Vol. 20, No. 2, 2017
Studiobinder Inc. [Internet]. Available: https://www.studiobinder.com/
S. Benini, L. Canini, and R. Leonardi, "Estimating Cinematographic Scene Depth in Movie Shots," in Proceedings of IEEE International Conference on Multimedia and Expo, pp. 855-860, July 2010.
S. B. Park and G. S. Jo, "Role Grades Classification and Community Clustering at Character-net," The Journal of the Korea society of computer and information, Vol.14, No.11, pp. 169-178, 2009.
H. S. Kim, N. H and S. J. Kim, "Web Crawler Service Implementation for Information Retrieval based on Big Data Analysis," The Journal of Korea Digital Contents Society, Vol.18, no.5, pp. 933-942, 2017.
J. S Kim, C. J. Nan, B. T. Zhang, "Deep Learning-based Video Analysis Techniques," The Journal of the Korea Information Science Society, Vol. 33, No. 9, pp. 21-31, 2015.
S. I. Son, C. K. Park, "Design of Big Data Preference Analysis System," The Journal of Korea Multimedia Society, Vol. 17, No.11, pp. 1286-1295, 2014.
R. B. Girshick, J. Donahue, T. Darrell, and J.Malik, "Rich feature hierarchies for accurate object detection and semantic segmentation," in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 580-587, 2014.
Bum-Soon Hwang, Eui-Nam Huh, Sun-Moo Kang "Real-time log analysis system using Tensorflow," in Proceeding of the Korea Information Science Society, pp. 672-674, July 2017.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.