Title: 
Interactive Suggestion for Improving the Quality in Photo/Video Shooting

Speaker:
YOSHITAKA, Atsuo (JAIST)

Abstract: 
When we focus on the 'quality' of photo or video, two aspects of quality are considerable. One is coding quality, and the other is aesthetic quality. The former refers to resolution, number of quantization bits for color channel, and the latter refers to the 'appearance' of contents from the point of view of composition, depth of field, camera angle and shot size. In case of video, camera work such as zoom-in is also one of the elements of the latter. The former largely depends on the development of hardware devices. However, the latter is mainly due to the effort by a user who is manipulating a camera. Therefore, one of the issues of improving the aesthetic quality of a shot is how a still/video camera can properly assist the user for manipulation.

This talk concentrates on the issue of interactive assistance for improving the aesthetic quality of photo/video shooting, and will present some of our recent studies as the solution of this issue. Here, novice users, who do not have sufficient knowledge and/or techniques for taking desirable shots, are assumed as the target users.

First, a method for suggesting composition will be taken into discussion. According to our survey of professional photos, there are typical types of composition which are often selected. The method for suggesting possible composition is to find primary subject and split lines in the background and find overlaps between possible choice(s) of composition so that it follows the preference by professionals.

Second topic focuses on the issue of suggesting the adjustment of the depth of field. Adjusting the depth of field is often taken into consideration in order to make a primary photographic subject clear, by making background blurred but it is in focus. However, improper adjustment of the depth of field may disturb a viewer to catch the surrounding environment or atmosphere. We will show a solution for this issue by adjusting focal length based on the relationship between the degree of blur and the size of primary photographic subject,  following the favor of professionals.

Lastly, the issue of suggesting desirable camerawork for the expression of nonverbal information is discussed. Shot size, camera angle, and zoom-in/out operation are known as factors for affecting visual expression of nonverbal information such as strength, weakness, strain, or desolation. A video camera with sensors to recognize shooting operation is introduced, which suggests proper shot size, camera angle and/or zooming operation as a solution of this issue.