erates the shown phosphor widgets—using the mouse or a shortcut—a stylized afterglow effect shows how they have changed. In some cases, such as the slider, the afterglow is an almost realistic depiction of the motion that took place during the change. Widgets that change in more complex ways, such as the combo box, are provided with a more abstract and symbolic afterglow in order to limit clutter.
Phosphor widgets are designed to focus users’ attention on objects that have changed. If users believe a manipulation took place in error, they can undo the action by returning the widget to the previous state suggested by the afterglow. Afterglow also helps users to observe changes that they have not initiated and hence provides better understanding in scenarios such as remote collaboration.
Afterglow effects are typically set to fade over a period of a few seconds. Actions occurring in rapid succession will therefore result in multiple concurrent afterglow effects. This is intended to help users catch up with fast bursts of activity as might occur during demonstrations or collabora- tive work.
Resulting benefits The proposed approach differs substantially from animated transitions: animated transitions explain the transition and then continue the regular execution of the program; phos- phor transitions do both at the same time.
This parallelism results in three main benefits: (1) Users can choose whether to attend to the explanation or to con- tinue with the regular program execution. Users are never forced to wait. (2) Since users are never forced to wait, ad- ditional display time comes at a low price. Inexperienced or distracted users can therefore be accommodated with in- creased afterglow durations. (3) Since display time comes at a low price, application designers can pick a reasonable upper bound. This frees them from having to hand-optimize duration—a major challenge faced by designers of ani- mated transitions.
The use of phosphor widgets introduces a tension between screen real estate and interaction time. Because phosphor widgets are susceptible to clutter, they require careful de- sign.
In the remainder of this paper, we give a brief overview of the related work. Then we take a closer look at the “visual language” of phosphor. We present designs for transitions for different types of interface objects and explain how to minimize clutter. After a brief description of our implemen- tations we present two user studies. The first study finds significant performance benefits for phosphor over a con- trol condition: Participants performed a simulated collabo- ration task faster when widgets were provided with an af- terglow. The second user study finds that phosphor’s task performance is similar or better to animated transitions. We conclude with a summary of our findings and an outlook to future work.
RELATED WORK Two main fields of related work for Phosphor are animated transitions and diagrammatic explanations.
Animated Transitions Animated transitions are one of the eight classes of anima- tion in the user interface . Benefits of animated transi- tions include that they can help increase the saliency of no- tifications , draw attention to peripheral displays, such as stock tickers , and that they can help illustrate causal relationships . Animated transitions can help users fol- low transitions between views , e.g., in applications dis- playing complex data, such as trees . By adding effects inspired by cartoons such as anticipation and follow- through, researchers have obtained a more lifelike effect (cartoon animation [9, 31]).
Research has not converged on consistent results regarding the efficiency of animated displays . Animated illustra- tions may require more cognitive load than static ones . Psychophysics research has shown that most users have difficulty tracking five or more objects [8, 36, 25]. Motion is hard to ignore and may thus cause users to be distracted by animated transitions .
Stasko  points out that animation duration is a crucial factor in the design of animation. To minimize lag, an ani- mation should be fast; making an animation too fast, how- ever, may lose the user. Researchers exploring animation durations have found that 300ms can work well for simple scrolling transition , while comprehending 3D transi- tions can require several seconds . Optimum animation speed depends on user- and situation-specific factors such as familiarity, expectation, attentiveness, and perceptual abilities and therefore are difficult to predict.
While designers of phosphor objects also need to set the duration for fading the afterglow, the question is less crucial because the afterglow does not prevent users from continu- ing their task.
Diagrams in information visualization In the fields of visualization and graphics, researchers have proposed illustrating dynamic phenomena using static de- pictions. Diagrammatic illustrations are amenable to print- ing , and can help users discover trends in large sets of motion data . On the flip side, users do not process dia- grams immediately and as a whole; users first have to dis- cover the best order to process the information .
Diagrammatic summaries come in many different styles. Feiner borrows principles from technical illustration , while Hill and Hollan use them to illustrate the past usage of a document . Carefully selected individual frames can be combined to form a strobe effect (Action Synopsis ). Chronovolumes combine successive frames into a con- tinuous motion blur. They use color transitions to depict the progression of time . Speed lines  are a more ab- stract type of motion blur created using non-photorealistic rendering . Speed lines have also been used to enhance the experience of animation sequences in video games  and to help users make sense of game map overviews .