SMPTE Timecode in the DTV Era
Television engineers have been using SMPTE timecode for about three decades now. It has facilitated electronic editing and the synchronization of a variety of video devices, to name just two benefits, and it would be impossible to do television as we know it today without it.
When SMPTE timecode was invented, we television engineers lived in a world that dealt with only four video temporal standards or, as we more commonly know them, frame rates.
Monochrome NTSC ran at a rate of exactly 30 frames per second, and so did 1125/60 analog HDTV and 30 fps film. Color NTSC runs at a rate of 30 frames per second multiplied by 1/1.001, or approximately 29.97 fps.
It should be noted parenthetically that all modern NTSC television equipment runs at the 29.97 fps rate and utilizes color burst, whether the video is color or monochrome; 525/60, 601 component SD video may run at either 29.97 or 30 fps. PAL and 625/50 component video run at exactly 25 frames per second. Finally, the preponderance of film is shot at 24 frames per second.
Because SMPTE timecode was invented to work in the world described above, it addresses just four frame rates: 30, 29.97, 25 and 24 fps.
The DTV world we live in today employs a few additional frame rates, for which there is no timecode. For example, one of the HDTV scanning formats that is now broadcast virtually on a daily basis is 720p/60 (really 720p/59.94, of course, as it must be synchronized to an NTSC-based plant), which runs at a rate of –approximately – 59.94 frames per second. There is no 60 fps timecode, but before we hastily telephone SMPTE, it should be pointed out that – as we will see later – this is not of great consequence in practical terms.
Timecode used in television – i.e., 30, 29.97 and 25 fps – may be one of two varieties: linear timecode or, as it was formerly known, longitudinal timecode (LTC) and vertical interval timecode (VITC).
Longitudinal timecode is a digital signal that is modulated using frequency shift keying (FSK) onto a carrier in the audio frequency range.
This type of timecode may be recorded on analog audio tracks and was called longitudinal timecode because it was recorded on the longitudinal or linear (as opposed to FM or some other technology) audio tracks of video tape recorders. It may also be distributed throughout a television plant on twisted-pair cables and otherwise handled like an analog audio signal.
Vertical interval timecode is a digital signal modulated onto lines of the vertical blanking interval of the video signal.
When comparing LTC and VITC, we find similarities as well as differences. LTC is a slower signal; an LTC "word" is about the length of a video frame (or about 1/30 second in NTSC), while a VITC timecode "word" is contained in a single interlaced field (or equivalent to about 1/60 second in NTSC). Although a VITC word is about half the temporal length of an LTC word, it contains about the same amount of information.
Because LTC is a slower signal, it may be successfully read at faster tape shuttle speeds than VITC, but only VITC can be read when the tape is not moving.
The smallest unit of resolution in LTC is a frame but because a VITC word is a half-frame in duration, VITC can differentiate fields – as observed when a tape is jogged field-by-field on a VTR while observing the burned-in timecode on the monitor screen. We will see, for example, frame 01, followed by frame :01 with some designator such as an asterisk. The asterisk or other such delimiter indicates that we are seeing field 2 of a given frame.
TO DROP OR NOT TO DROP
The 29.97 fps timecode itself comes in two varieties: dropframe and nondropframe. Dropframe timecode, of course, does not imply that any video frames are dropped. Timecode is fundamentally a way to label frames to keep track of them and either drop or nondrop timecode will uniquely identify each frame. However, if 29.97 fps video is labeled with nondrop timecode, its indicated timings will not accurately correspond with real time.
Because 29.97 fps video is running slower by a factor of 1/1.001 than 30.00 fps video, it will contain 30 x 1/1.001 or 0.03 extra fps, or 108 extra frames, corresponding to 3.6 extra seconds per hour.
If nondrop timecode is applied to 29.97 fps video, an indicated hour of program material in fact runs 59 minutes and 56.4 seconds. If a 24-hour television station were to time all its material using nondrop timecode, it would end the day having to fill an extra 86.5 seconds with color bars!
The solution to this problem is dropframe timecode, which permits timing 59.94 fps video accurately with real time by dropping the extra frames in the count. In order to distribute the dropframe effect evenly across each hour, the count drops two frames each minute except for each tenth minute – that is, no frames are dropped at minutes 00, 10, 20, etc.
To give an example, when the dropframe timecode count reaches 1:00:59:29, the next count is not 1:01:00:00, but rather is 1:01:00:02, with frames 00 and 01 being skipped. When this is done, program material that is indicated by the timecode to be one hour in duration is, in fact, exactly one hour long. Production and postproduction people hate to deal with dropframe timecode, but we who live in the 29.97 fps world rely on it and will as long as we must run our plants at the NTSC frame rate.
CHANGES DUE TO DIGITAL
Two issues have arisen as a result of new practices brought about by DTV. The first is that there is no 60 fps timecode. This is of little consequence in practice, as the field identification facility in VITC may be (and is) used to identify individual frames in 60 (or 59.94) fps video. Thus, the same editing approaches used for 60-frame-per-second-interlaced video may be used for 60-frame-per-second progressive video.
The second issue arises because of the increasingly practiced recording of 24-frame (or 23.98-frame) video. When 24-frame film is transferred to 29.97 fps interlaced or 59.94 fps progressive video, it is slowed down by the same factor, running at 24 x 1/1.001, or about 23.98 fps.
In recent years, the practice of mastering to 24 (23.98) fps video, rather than 29.97 fps or 59.94 fps video, has increasingly been used. The reasons for this are apparent, in that it establishes a one-to-one correspondence between film frames and video frames. Further, the practice of shooting on 24 fps video rather than on 24 fps film has been gaining some currency in the production community.
Although nominal 23.98 fps video is being recorded, there is no 23.98 fps dropframe timecode. This has generated some process problems where conversions are required between 24 fps (nondrop) timecode and dropframe 60 Hz-based timecode.
There are efforts under way to rectify these problems, both on the standards front and on the translation-software front. This is just another facet of the "interesting times" in which we television engineers live.
The latest product and technology information
Future US's leading brands bring the most important, up-to-date information right to your inbox