MixT: automatic generation of step-by-step mixed media tutorials

As software interfaces become more complicated, users rely on tutorials to learn, creating an increasing demand for effective tutorials. Existing tutorials, however, are limited in their presentation: Static step-by-step tutorials are easy to scan but hard to create and don't always give all of the necessary information for how to accomplish a step. In contrast, video tutorials provide very detailed information and are easy to create, but they are hard to scan as the video-player timeline does not give an overview of the entire task. We present MixT, which automatically generates mixed media tutorials that combine the strengths of these tutorial types. MixT tutorials include step-by-step text descriptions and images that are easy to scan and short videos for each step that provide additional context and detail as needed. We ground our design in a formative study that shows that mixed-media tutorials outperform both static and video tutorials.