Comparison of Photo Library Programs

Introduction

Technology advances in digital cameras, scanners, and electronic storage devices has made digital photography a much more attractive method of keeping photos. However, digitizing photos is only part of the process. Photo library software is needed to provide organization, presentation, and navigation of large personal photo libraries. In an experiment we compared three photo library software packages to find out how easy it was to locate pictures within their photo libraries. For this experiment we defined libraries as user defined groups of photo collections, which in turn contain groups of photos. The experiment measured how long it took users to find given photos within photo libraries of varied sizes and between different photo collections.

We compared two of the more popular photo library software packages, PhotoSuite III and ACDSee, with a version of the University of Maryland Human-Computer Interactions Laboratory's (HCIL) PhotoFinder project. The three software packages all provide users with the ability to view their collections as a group of thumbnails. However, they differed in the way the managed the thumbnail view and the way they provided collection navigation.

ACDSee uses the windows directory structure to manage collections. To navigate among collections, a Windows Explorer style navigation frame is supplied on the left-hand side. The images within a directory are shown as thumbnails in the right hand frame. A larger version of the currently selected photo appears in the center frame. The ACDSee software package basically provides a Windows Explorer interface to photo library browsing, searching, and sorting.

The newest version of MGI's PhotoSuite software package provides a nice slick looking interface for browsing and managing photo collections. It provides a full size picture of the selected image in the main center frame. Thumbnails are provided in a scrolling bar on the side that allows users to select the displayed image. In addition to this view another thumbnail view is available that uses the main center frame to display thumbnails, thereby allowing more thumbnails to be viewed at once.

PhotoFinder is a part of an ongoing HCIL project to create a photo library software package that allows users to directly annotate their photo libraries and then use that annotation to search collections. The interface consists of three main frames: the library viewer, the collection viewer, and the photo viewer. The library viewer is the left-hand frame and displays stacks of photos that represent the different collections within the library. The collection viewer is the upper right-hand frame and displays thumbnails of the photos contained in the selected collection. The photo viewer is the lower right-hand frame and displays an enlarged view of the currently selected photo.

The experiment focuses on finding out how the differences between the interfaces affect the time it takes users to navigate and browse the libraries and how the users subjectively judge the different interfaces. For instance, will those users familiar with Windows prefer the ACDSee Windows-style interface? How will the different methods of thumbnail display affect users seek time? Does the arrangement of the different navigational windows make a difference in how users look for an image? For larger collections, should thumbnails be shrunk to fit more on the screen at once?

We hypothesize that users will find it easier to navigate and browse through PhotoFinder, rather than PhotoSuite III or ACDSee. Users will be able to finish tasks more quickly using PhotoFinder. We feel that users will prefer PhotoFinder over the other two softwares because PhotoFinder consists of three frames: the library viewer, the collection viewer, and the photo viewer. Users will not get lost in the library, whereas in PhotoSuite III, users may forget which collection they are in.

In the experiment we tested subjects on the amount of time it takes them to perform five tasks using one of the photo library packages. The tasks ask subjects to find photos in collections of varying sizes. For some tasks the picture was shown to the subjects and others it was just described. Subjects also had to figure out where a picture is located among several collections. In addition, subjects filled out background and subjective satisfaction surveys.

Review of Previous Experiments

Some of the references we found contained experiments done on digital photo libraries. The experiments we will review are "FotoFile: A Consumer Multimedia Organization and Retrieval System" and "PhotoFinder: Browsing and Searching for Personal Photos within a Collection."

In the FotoFile article, Hewlett Packard found that one of the sources of frustration for consumers was the organization and retrieval of digital images. Organizing and managing photos was a tedious task and it was error prone (Kuchinsky, et al., 1999). The FotoFile article brought up the point that attention has been directed to the task of "direct search." Direct search is a search in which the user has a goal in mind. Little attention was directed at browsing through collections with no goal in mind. The Hewlett Packard research team felt that more research should be done on browsing.

In the FotoFile experiment, the research team set up focus groups in Denver and San Francisco. There were different types of focus groups - business and home participants. In the focus groups, issues of organizing, searching, and sharing photos were discussed. Keyword-based searches and automated searches were discussed. Each photo has a text description attached to it. Keyword-based searches search through the text descriptions. Automated searches are searches that use visual attributes of images, such as color. The research team found that:

• Home participants easily grasped the concept of keyword-based searching.
• Both groups grasped the concept of automated indexing and saw its benefits.
• Home participants liked the idea of browsing, while business participants did not.

In conclusion of the FotoFile experiment, the research team built a system which balanced keyword-based searches with automated searches and which also balanced direct searching with browsing.

In fall of 1999, graduate students at the University of Maryland implemented a digital photo library, PhotoFinder, and did a study on it. These students searched through other digital photo libraries and found that they were difficult to use, and the searching and browsing capabilities were not efficient (Kang & Tong, 1999).

The students used four interface strategies which include Direct Manipulation and Dynamic Query, Visual Information Seeking Strategy, Query Previews, and Multiple Tightly Coupled Views. For Direct Manipulation and Dynamic Query, the students wanted the interface to "produce immediate, continuous and visual feedback of search/query results" (Kang & Tong, 1999). For Query Previews, the students used dynamic query techniques.

For Visual Information Seeking Strategy and Multiple Tightly Coupled Views, the students built a three-level hierarchy, and all three levels were coupled together. The students split PhotoFinder into three major parts: Searching Tool Window, Thumbnails Browser Window, and Detail Viewer Window. In the Searching Tool Window, users are allowed to make queries in order to search for photos. The Thumbnails Browser Window allows users to view all of the thumbnails of a collection of photos. The Detail Viewer Window allows users to view one photo in a larger size.

In conclusion of the PhotoFinder experiment, the students felt that PhotoFinder provided "better searching and browsing capabilities" (Kang & Tong, 1999). The students felt that PhotoFinder was a "powerful photo-browsing tool."

Relevant Theories

Using the GOMS model, performance might vary between the three software packages. The GOMS model "postulates that users formulate goals and subgoals … which they achieve by using methods or procedures" (Shneiderman, 1998, p.55). The goal of every user in our experiment will be to search for photos.

Mouse clicks to switch between thumbnails and photos will affect performance. In PhotoFinder, there is a thumbnail viewer and a photo viewer shown on the screen at the same time. This may minimize the amount of switching between thumbnails and photos, since users only need to click the thumbnail once in order for it to appear in the photo viewer window. In ACDSee, there are also two separate windows for thumbnails and photos, so mouse clicks may be kept at a minimum as well. In PhotoSuite III, the task is more difficult. If users click on the thumbnail, the thumbnails move to the left frame, while the photo appears on the main frame. If the photo is not one which the users want, they may have to either scroll in the left frame (which may not be practical for large collections since photos are shown in one column), or click on the album name in order to bring the thumbnails back to the main window. The problem is that the scroll bar returns to the top of the collection, and the users may lose their place. Users not only have to click and scroll, but time is affected as well.

For our largest task, users will have to use the scroll bar. This may affect the performance times as well. If some users are more comfortable with the "slide show" feature, they may choose to use this feature. This would minimize the mouse clicks between thumbnails and photos. Instead, the user would be able to view each photo in a large size with only one keystroke (or mouse click) to switch between photos.

Considering the amount of operations in each package needed to complete a task, PhotoFinder may require the least. Users may prefer PhotoFinder because of the less time it takes during browsing, and the least amount of operations.

Department of Computer Science: Direct questions and comments to the student editorial team