Open topic with navigation
Time mode creates documents that represent a fixed amount of video content. A document contains all of the records that occurred during the time interval, or that overlap the start or end of the interval. You can define the duration of a document, for example 30 seconds. The duration is measured using video time, so you do not need to modify the duration if the ingest rate is slower or faster than normal (playback) speed.
Time-based output does not mean that Media Server outputs data at regular intervals. Media Server cannot output data for a video segment until all records in that segment have finished. As a result, Media Server might produce several documents, which represent consecutive time periods, at the same time.
For example, if you run logo detection on a news broadcast, the news logo could be present on screen continuously for an hour. In that case, Media Server does not output any data until the logo disappears. Although Media Server creates documents that each represent 30 seconds of content, all of the documents are output at the end of the hour when the record describing the logo finishes.
In time mode, all records are output to at least one document. If a record spans more than one interval it is output to multiple documents.
Consider these records. The following diagrams demonstrate how Media Server constructs documents in time mode.
The first document contains the records related to the first interval (in this case, 30 seconds) of video content. Media Server does not output data at the 30-second point because a face recognition record has not ended (the face is still present in the video). Media Server can output data about the first interval as soon as this record ends.
The second document contains the records related to the second interval (in this case, 30 seconds) of video content. Notice that the second face recognition result is output in the second document as well, because it relates to both time intervals.
The third document contains the records related to the third interval (in this case, 30 seconds) of video content.
Time mode is simple to set up but documents might begin and end in the middle of a sentence or segment, rather than at meaningful point in the video.