NudgeCam: Intelligent capture

It can be difficult to shoot high quality video with your mobile phone. Some online tools, such as YouTube, can help correct common problems such as jitter and lighting in post processing. But of course these tools cannot add back content that you never shot in the first place. NudgeCam is a mobile application I built that helps users make sure they capture the content they mean to. NudgeCam provides real time feedback to make sure that faces stay in frame, audio quality is good, and panning motions are not too violent. The goal is to make sure that captured video content is good enough so that post processing methods can be effective.

NudgeCam also supports guided capture. This is particularly useful when an expert (e.g., senior technician at a central office) needs to gather data from people in the field. A template creator can show users media that demonstrates relevant content and can tell users what content should be present in each captured media using tags and other meta-data such as location and camera orientation.

NudgeCam is a (multi-threaded) Android application. It runs entirely on the mobile device with no backend support.

1

NudgeCam integrates well known capture heuristics, including interview guidelines. When a face is detected that is too small the face is outlined in red with markers at the corners suggesting that the videographer increase the size of the face in the view.

2

NudgeCam monitors other data streams as well. Here, it notifies the user when no audio has been detected (yellow), audio has been detected and the quality is good (blue), or audio has been detected but the quality is poor (red).

Reminders also appear in an overlay during recording to remind the videographer of suggested video content. The videographer can check off reminders similar to a to-do list. When a topic is clicked it changes color and its completion time is written to local storage.

3

Users can playback recorded video on the mobile device. The player is augmented with bookmarks indicating the times at which each topic appears (e.g., the time in the recording when the user checked the topic off the to-do list).