Janks (dropped or delayed frames) are still common in everyday smart phone animations, despite advances in mobile processing power. The goal of this project is to identify and characterize the properties of jank that most affect user experience. To achieve that, we divided this project into of two parts: understanding janks and quantifying the user perception of janks.
PART 1: Understanding Janks with Computer Vision
To understand the properties of janks, we developed a custom computer vision program that analyzed hours of real-world smart phone interactions. We studied the properties of jank during user interaction by examining the videos using a custom analysis program.
PART 2: User Perception
Through an in-the-wild experiment covering over 3900 participants, we gathered subjective user feedback on four common interaction scenarios where controlled janks were inserted. We analyzed the effect of different factors on jank perception and acquired a holistic model to quantify the subjective perception of janks.
The results from this study can be useful for analyzing and improving user experience of animations in mobile applications and operating systems.
- (In preparation for ACM TOCHI)
- Developed the video analysis platform for janks in transition animations using computer vision
- Designed an Android application that inserted janks during user interaction in four common scenarios and gathered user feedback
- Conducted the large-scale in-the-wild user experiment
- Analyzed the experiment data to acquire the perception model of janks
- Working on a paper for ACM TOCHI (as the first author)