Starts the generation of the video summary, so it can be retrieved later using the returned job identifier. A video summary is generated using the smart detection snapshots within the time period. The API will use all available snapshots to generate a video, however when too many snapshots are available, an uniformly interleaved selection of the available snapshots is chosen to generate the video. When the number of snapshots is low, the speed and length of the video will gradually decrease. No video summary will be generated when there are only very few snapshots available.