Image Size vs. Scrolling in Photo Thumbnail Browsers

Conclusions

Impact for practitioners

The results of this experiment suggest some useful guidelines for designers of thumbnail browsers. We found that fitting thumbnails in a single window allowed for faster searches up to a certain number of images, but that after this point, performance time suffered as thumbnails became too small to see clearly. The scrolling required for viewing larger thumbnails with 48 images slowed people down about the same amount as looking at very small thumbnails, but people preferred scrolling in this case. This suggests three possible design strategies. First, designers could vary thumbnail size with collection size, using larger thumbnails for smaller collections and smaller (but not too small) thumbnails for larger collections. This design minimizes wasted space in the window with small collections and minimizes scrolling for large collections. A second design choice would be to pick a single thumbnail size that is as small as possible to minimize scrolling but not so small that it is hard to see. This second solution may waste window space with small collection sizes by not filling it completely, but has the advantage of using consistently sized thumbnails. A third possibility would be to allow users to customize the thumbnail browser interface, selecting a particular thumbnail size for each different situation.

Suggestions for future researchers

An interesting extension to our experiment would be to experiment with users over a longer term to see if their preferences change with extended usage. We hypothesize that as users become more familiar with the interface and spend more time using it, their preference may change to the interface in which they have a better performance time. In particular, we would expect to see users change their preference in the collection of size 36 from the fixed treatment to the variable treatment. As our experiment focused on users who were not familiar with the photographs they were identifying, it would be interesting to experiment with users who are familiar with the photographs, since this may be a more common use situation. We hypothesize that in this case, smaller thumbnails might be ok to use because of users' familiarity with the photographs. Another useful experiment would be to see if varying thumbnail sizes based on collection sizes is disruptive to users. We hypothesize that users might prefer that thumbnails remain the same size regardless of collection size, even if this means some wasted space for very small collections, and some scrolling for very large collections. Finally, it might be useful to study online photo thumbnail browsing applications to see if factors like network connection speed influence design choices.

Refined theory

Our results showed that scrolling in thumbnail image browsers never improved task time completion. However, in larger collections, as thumbnails became smaller with a non-scrolling design, users preferred scrolling with larger images. For users not familiar with the images they are viewing, thumbnails of 80 x 80 pixels are probably too small to identify easily in a thumbnail browser, at least with moitor settings similar to the ones tested. Given a choice between a design that requires users to view images of this size or smaller and a design that requires more scrolling with larger images, the latter is probably preferable.

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