The current Visual Explorer represents each television news broadcast using a single frame extracted exactly 60 seconds into the broadcast. For many broadcasts, this yields a perfectly reasonable visual representation of the broadcast. However, for many, it produces an unrelated image, such as when a broadcast is opened with a stock intro sequence. For example, a CSPAN broadcast of a Pentagon briefing from last month is currently represented by the generic stock image below because it opens with a stock sequence lasting a full minute:

In contrast, our new algorithmic thumbnail generator, which uses ffmpeg's representative frame extractor from the first 6 minutes of the broadcast, yields the image below as the representative thumbnail of the broadcast, which accurately captures the focus of the broadcast, demonstrating the power of this new approach: