PARR_A New System for Creative Photography/ GCT 503 Final Project / Fall 2010

PARR

A New System for Creative Photography

Bongjun KIM, Ikdu KIM, Kayyoun AHN, Kyehyun KIM and Yong Hun KIM
final project | GCT503 | Fall 2010
y@kimyonghun.com

1 Introduction

Approaching computational photography from a creative perspective, we invented a new photographic system named 'PARR' for extending the photographers' ability of expression. The system observes the context of the environment with various sensors attached to computer. And it chooses the 'decisive moment' based on analyzed quantitative data through data processing technology.

2 Background

Computational photography means sensing strategies and algorithmic techniques that enhance or extend the capabilities of digital photography. The output through computational photography is not always special photograph, but it that could not have been taken by an ordinary camera. In this area, high dynamic range imaging, flash-no flash imaging, coded aperture and coded exposure imaging, photography under structured illumination, multi-perspective and panoramic stitching, digital photomontage, all-focus imaging, and light field imaging are the representative technology in nowadays. [1]

For maximizing the use of algorithmic techniques and widening sensor adaptabiltiy, we decided to build a photographic system rather than a camera, which will shift phtographers' process of taking photos. The system's main architecture is the computer connected with sensors and image devices.

3 System Model
3.1 Overview

PARR, the system we developed is constituted of the computer connected with sensors and image devices. The system makes the most of computer's data processing performance which complements photographers' decision making. And it is open to adaptation of various sensors which extends photographers' sensing abiltiy beyond human capability. In addition, image devices like digital cameras can be attached to the system for producing higher quality images. Figure 1 is the concept of the system.

Figure 1: The overview of the system 'PARR'

3.2 Style

Our project is named after British photographer Martin Parr, one of the most successful photojournalists who became a member of Magnum Photos in 1994[2].

Martin Parr examines issues concerning mass tourism, consumerism, global society and culture, and makes cynical comments on them with serial photos[3].

Parr's hyaline realism comes from his dry point of view, and also by grasping things near us. It is not hard to find cynicism in his photos. Many pundits say Parr's series show typical English dry sense of humor. We figured that the view in Parr's work is rather mechanical than cynical. Art critic Val Williams commented that Parr's work shows "an exercise in looking"[4].

Our system 'PARR' mimics Martin Parr's photographic style. It observes the context of the environment with a mechanized view point. The gathered data from sensors gets analyzed by the processing algorithms, which would decide the capturing moment of the scene. And the system repeats the designated performance to produce a series of images.

3.3 Technical Features

Figure 2: GUI of PARR consists of a main window showing live feedback, result window showing last taken picture and slider bars to manipulate feature attributes.

Our program currently provides the following features.

-People detection
It is done with HOG algorithm provided by OpenCV. It analyzes the image inputs from the web-cam and tags it with box information that specifies the locations and sizes of all the people detected.

-Color detection
It detects the major color inside a user specified region. The color at the center of the region is heavily considered compared to the color at the boundary of the region. It also takes images as input and tags them with the region and color information.

-People detection with color
The two features above can be combined to detect people wearing clothes of a certain color. First, people are detected in the image inputs and the region for color detection becomes the box information from the people detection. This feature tags the image input with the box information from people detection and the color information from color detection.

-Sound detection
It detects the decibel of the input audio and tags them with true/false information depending on the user specified value.

4 Experiments
4.1 Pilot Study

We conducted several test shots for a pilot study. Throughout this process we refined the system by shortening detection time and enhencing user interface.

And also the result of the test shots convinced us that the system has a possiblity to be a new image producing system.

Figure 3: Desktop computer installed on the 3rd floor, webcam(with USB extension cable) attached on the wall of 1st floor and digital camera installed by the window of the 3rd floor of a building.

Figure 4: Result photographs (Running human recognition function, the webcam recognizes passers-by and the digital camera takes photos.)

4.2 Workshop

On Dec. 4. 2010, we held a workshop in Hong-Ik university with several photographers who have years of experience in photograph. We provided them with the system program file and made test shots together. After interviewing the participants, we were informed that the system has advantages such as automation of taking photos and readiness of acquiring unusual angle of view.

4.3 Exhibition

We asked the photographers attended the workshop to produce images with the system for ten days. And we organized an exhibition titled 'SEEN' to demonstrate how creatively the system can be used and see audience's response to the photographs made by the system. The exhibition opened at corridor of KAIST N8, on Dec. 15. 2010. Many audiences said the photos were showing a strange aspects of the world.

Figure 5: Audiences at the exhibition

Figure 6: Installation view of the exhibition

See more works: 100 White Cars, 100 Screams

5 Conclusion

We believe that our system provides new possibilities of producing photographic images. The main advantage of the system is that it enables photographers to have extended senses by using sensors measuring physical quantity of the environment beyond human capability. And the measured physical property gets converted into signals to be processed to choose the decisive moment to capture the image on behalf of photographers. In addition, the automated system could produce images without the presence of photographers. It means that photographers can install the system more than one place to make pictures of multiple locations at the same time, and that they can control the system and work without limitation of space by utilizing network technology.

Future computers and sensors might have even more advanced capabilities and become more compact, which will enhance our system's processing efficiency, sensing capability and mobility. The system might someday be an independent entity producing images based on it own senses and decision. Then we might be able to see the yet unseen moments of our world.

See the making film of the project.

6 References

[1] Andrew Adams, Eino-Ville Talvala, Sung Hee Park, David E. Jacobs, Boris Ajdin, Natasha Gelfand, Jennifer Dolson1 Daniel Vaquero, Jongmin Baek, Marius Tico, Hendrik P. A. Lensch, Wojciech Matusik, Kari Pulli, Mark Horowitz and Marc Levoy, ¡°The Frankencamera: An Experimental Platform for Computational Photography,¡± ACM Transaction on Graphics, vol. 29, no. 4, 2010.

[2] Robert Ayers,¡± Martin Parr,¡± ARTINFO, Nov. 2006, http://www.artinfo.com/ news/story/24111/martin-parr/, retrieved 2008-07-14.

[3] Nathalie Johnson, ¡±The Photographer and The Photographed: The balance of power between a photographer and subject,¡± Ba (Hons) Documentary and Fine Art Photography, Stockport College., 2002.

[4] Val Williams, ¡°in Martin Parr,¡± London: Phaidon Press Ltd., p. 161, 2002.


		PARR
		A New System for Creative Photography Bongjun KIM, Ikdu KIM, Kayyoun AHN, Kyehyun KIM and Yong Hun KIM final project \| GCT503 \| Fall 2010 y@kimyonghun.com

1 Introduction		Approaching computational photography from a creative perspective, we invented a new photographic system named 'PARR' for extending the photographers' ability of expression. The system observes the context of the environment with various sensors attached to computer. And it chooses the 'decisive moment' based on analyzed quantitative data through data processing technology.
2 Background		Computational photography means sensing strategies and algorithmic techniques that enhance or extend the capabilities of digital photography. The output through computational photography is not always special photograph, but it that could not have been taken by an ordinary camera. In this area, high dynamic range imaging, flash-no flash imaging, coded aperture and coded exposure imaging, photography under structured illumination, multi-perspective and panoramic stitching, digital photomontage, all-focus imaging, and light field imaging are the representative technology in nowadays. [1] For maximizing the use of algorithmic techniques and widening sensor adaptabiltiy, we decided to build a photographic system rather than a camera, which will shift phtographers' process of taking photos. The system's main architecture is the computer connected with sensors and image devices.
3 System Model 3.1 Overview		PARR, the system we developed is constituted of the computer connected with sensors and image devices. The system makes the most of computer's data processing performance which complements photographers' decision making. And it is open to adaptation of various sensors which extends photographers' sensing abiltiy beyond human capability. In addition, image devices like digital cameras can be attached to the system for producing higher quality images. Figure 1 is the concept of the system.
3 System Model 3.1 Overview		Figure 1: The overview of the system 'PARR'
3.2 Style		Our project is named after British photographer Martin Parr, one of the most successful photojournalists who became a member of Magnum Photos in 1994[2]. Martin Parr examines issues concerning mass tourism, consumerism, global society and culture, and makes cynical comments on them with serial photos[3]. Parr's hyaline realism comes from his dry point of view, and also by grasping things near us. It is not hard to find cynicism in his photos. Many pundits say Parr's series show typical English dry sense of humor. We figured that the view in Parr's work is rather mechanical than cynical. Art critic Val Williams commented that Parr's work shows "an exercise in looking"[4]. Our system 'PARR' mimics Martin Parr's photographic style. It observes the context of the environment with a mechanized view point. The gathered data from sensors gets analyzed by the processing algorithms, which would decide the capturing moment of the scene. And the system repeats the designated performance to produce a series of images.
3.3 Technical Features		Figure 2: GUI of PARR consists of a main window showing live feedback, result window showing last taken picture and slider bars to manipulate feature attributes.
3.3 Technical Features		Our program currently provides the following features. -People detection It is done with HOG algorithm provided by OpenCV. It analyzes the image inputs from the web-cam and tags it with box information that specifies the locations and sizes of all the people detected. -Color detection It detects the major color inside a user specified region. The color at the center of the region is heavily considered compared to the color at the boundary of the region. It also takes images as input and tags them with the region and color information. -People detection with color The two features above can be combined to detect people wearing clothes of a certain color. First, people are detected in the image inputs and the region for color detection becomes the box information from the people detection. This feature tags the image input with the box information from people detection and the color information from color detection. -Sound detection It detects the decibel of the input audio and tags them with true/false information depending on the user specified value.
4 Experiments 4.1 Pilot Study		We conducted several test shots for a pilot study. Throughout this process we refined the system by shortening detection time and enhencing user interface. And also the result of the test shots convinced us that the system has a possiblity to be a new image producing system.
4 Experiments 4.1 Pilot Study		Figure 3: Desktop computer installed on the 3rd floor, webcam(with USB extension cable) attached on the wall of 1st floor and digital camera installed by the window of the 3rd floor of a building. Figure 4: Result photographs (Running human recognition function, the webcam recognizes passers-by and the digital camera takes photos.)
4.2 Workshop		On Dec. 4. 2010, we held a workshop in Hong-Ik university with several photographers who have years of experience in photograph. We provided them with the system program file and made test shots together. After interviewing the participants, we were informed that the system has advantages such as automation of taking photos and readiness of acquiring unusual angle of view.
4.3 Exhibition		We asked the photographers attended the workshop to produce images with the system for ten days. And we organized an exhibition titled 'SEEN' to demonstrate how creatively the system can be used and see audience's response to the photographs made by the system. The exhibition opened at corridor of KAIST N8, on Dec. 15. 2010. Many audiences said the photos were showing a strange aspects of the world.
4.3 Exhibition		Figure 5: Audiences at the exhibition Figure 6: Installation view of the exhibition See more works: 100 White Cars, 100 Screams
5 Conclusion		We believe that our system provides new possibilities of producing photographic images. The main advantage of the system is that it enables photographers to have extended senses by using sensors measuring physical quantity of the environment beyond human capability. And the measured physical property gets converted into signals to be processed to choose the decisive moment to capture the image on behalf of photographers. In addition, the automated system could produce images without the presence of photographers. It means that photographers can install the system more than one place to make pictures of multiple locations at the same time, and that they can control the system and work without limitation of space by utilizing network technology. Future computers and sensors might have even more advanced capabilities and become more compact, which will enhance our system's processing efficiency, sensing capability and mobility. The system might someday be an independent entity producing images based on it own senses and decision. Then we might be able to see the yet unseen moments of our world.

		See the making film of the project.

6 References		[1] Andrew Adams, Eino-Ville Talvala, Sung Hee Park, David E. Jacobs, Boris Ajdin, Natasha Gelfand, Jennifer Dolson1 Daniel Vaquero, Jongmin Baek, Marius Tico, Hendrik P. A. Lensch, Wojciech Matusik, Kari Pulli, Mark Horowitz and Marc Levoy, ¡°The Frankencamera: An Experimental Platform for Computational Photography,¡± ACM Transaction on Graphics, vol. 29, no. 4, 2010. [2] Robert Ayers,¡± Martin Parr,¡± ARTINFO, Nov. 2006, http://www.artinfo.com/ news/story/24111/martin-parr/, retrieved 2008-07-14. [3] Nathalie Johnson, ¡±The Photographer and The Photographed: The balance of power between a photographer and subject,¡± Ba (Hons) Documentary and Fine Art Photography, Stockport College., 2002. [4] Val Williams, ¡°in Martin Parr,¡± London: Phaidon Press Ltd., p. 161, 2002.