A computer monitor and a television set are both raster-based display systems: an image is delivered to the screen line-by-line. Modern monitors typically function in a progressive scan mode, however, and good old-fashioned broadcast TV sets use interlaced scanning.
This document presents an overview of the features of interlaced video streams which are essential to understand for working with digital video. The following documents at Lurkertech present an even better account (with even better info-graphics), but may go into more detail than you want right now:
Progressive scanning is straightforward. Pictures are sent to the screen as a stream of frames. Each frame contains a complete screen-full of image data, with one scanline for every scanline on the display. The frame rate specifies the number of frames per second in the video stream.
The television set is an interlaced output device. Video is sent to the TV as a stream of fields. Each field contains only half of a screen's worth of image, because a single field contains data for every other scanline of the display. Half of the fields are upper fields, because they contain data for every other scanline, starting from the very top (and going to the second from the bottom). The other fields are the lower fields, which start at the second-from-the-top scanline, and end with the very last scanline.
The stream of fields to the TV alternates between the two types: ...upper, lower, upper, ... (There is no "starting field" here --- they just alternate, forever, in both directions, as far as the TV is concerned.) Two fields are required to draw in order to all of the scanlines in the screen. Since the two types of field mesh together on the screen, the video stream is said to be interlaced.
The field rate specifies the number of fields per second in the video stream. One also talks about the frame rate, which is half the field rate, since it takes two fields to draw a video screen. Technically, however, the television set itself knows nothing about frames. It just sees a stream of alternating upper and lower fields.
Keep in mind that two interlaced fields do not make a progressive frame! The scanlines of a single progressive frame are all sampled starting at the single point in time. Interlaced fields are sampled starting at different points in time. In the above figure, you can see that the red ball moves downward between fields. This is why interlaced and progressive streams are not interchangeable, and why "deinterlacers" exist.
Note also that a single progressive frame has better spatial resolution than a single interlaced field, since it has a full screen's worth of scan lines. However, an interlaced stream actually has better temporal resolution than a progressive stream, since it takes samples twice as often.
When an interlaced video stream goes through digital video editing equipment (and often analog as well), the fields are paired off into frames. The video stream is treated on a frame-by-frame basis from that point onward, until it finally gets shipped back out to a TV.
Each frame being shuffled through the editing process still contains an upper field and a lower field. The fields are typically presented in one of two ways. A field-sequential frame is encoded as two half-height images: all the scanlines for one field, followed by all the lines for the other field. An interleaved frame is encoded as a single full-height image: the scanlines from each field are placed in their proper spatial locations in the image.
field-sequential | interleaved | |
This only applies to interlaced streams. Progressive streams are already shipped as whole frames.
Note that when a stream of fields is divvied up into frames, a choice must be made: should each frame start with an upper field first, or with a lower field? This choice is called the field dominance of the video stream. The notion of field dominance is only applicable to an interlaced, field-based stream which has been packaged into frames.
Knowing the field dominance of your video stream is important, because this choice must be maintained throughout the processing of the signal. If multiple video sources are going to be spliced together, then they all must have the same field dominance. For any video source, you only get one chance to choose the field dominance, and that is when the stream enters the editing chain (e.g. when an analog video signal is captured by a framegrabber).
(It is possible to drop a field from the start of video sequence, and then reframe the remaining fields, thus switching the upper-lower parity, but you will lose a field and screw up your timecodes in doing so.)
Typically, the choice of "upper-field-first" or "lower-field-first" is pretty arbitrary, but there are a few factors that may influence the decision:
The field dominance is a necessary descriptive bit for a frame-based stream. If the stream is field-sequential, then you always know which field comes first (because it actually comes first), but you do not implicitly know if the first field is an upper or a lower. If the stream is interleaved, then you always know which field is which spatially (because they are already in the frame in the proper locations), but you do not implicitly know which field should be displayed first in time.
What are "even" and "odd" fields? That's crazy-talk. Even worse, it's really ambiguous and confusing. Don't use those words.
maximum impact research Digital Media Group <dmg at mir.com> |
Last modified: Mon Oct 29 23:12:13 EST 2001 |
©2001 Matthew Marjanovic.
This material may not be republished in any form
without express written consent of the author.