Digital Standards Conversion

content. In standard-definition television, one 625/50 picture is constituted by 720× 576 pixels (720 × 486 for 525/60 pictures). The continuous stream is trans- formed into a set of clearly distinguishable discrete values that can be individually generated, accessed, processed, and manipulated. Such a signal representation has facilitated the development and construction of a rich panoply of processing and special effects tools that were once hardly imaginable and even more diffi- cult to realize in the analog age and domain. Huge new vistas of artistic creativity have been opened.

5.4 Digital Standards Conversion

A good example of the capabilities of digital technologies is the standards con- version. As explained in Chapter 4, in spite of all endeavors it was not possible to achieve satisfactory analog standards conversion. The goal seemed suddenly near at hand when the first all-digital standards converter appeared on the market. And, in fact, very soon after a series of improvements, broadcasters had in their hands excellent, practical, and affordable digital standards converters.

Converting from one scanning standard to another, from 525 to 625, or from the standard definition to high definition, is a very complex operation requiring changes in the following:

• line length (for example the line duration in the 525 standard is 52.86µs while in the 625 standard that duration is 52µs only)

• number of lines (including their repositioning since the lines in two scanning systems have different geometrical positions on the screen)

• number of fields (a given number of fields means also a given temporal sampling, that is, the movement of an object in front of the camera will be portrayed with 50 pictures per second in one case and 60 pictures per second in the other)

Consequently, it is not possible to just “drop lines” and simply repeat twice some of the fields when passing from 625 to 525, or to “drop fields” and repeat some of the lines when going in the opposite direction. Such procedures would result in unacceptable picture degradations. The digital standards conversion is based on the application of a technique called interpolation—the computing of the position and “importance” of each pixel from a number of samples taken in its vicinity and then the re-creation of a new set of pixels whose geometrical position and number will correspond to the requirements of the selected output standard. The simplest part of the standards conversion process is the change in line length, the horizontal conversion. The physical length of one line depends on the

width of a screen, but in two scanning standards, the duration of that line is different, or, in digital terms, the number of active pixels in 625/50 and 525/60 standards is different. Therefore the horizontal conversion consists of taking the active pixels from a line in one standard, computing from them a new set of pixels corresponding to the other standard, and interpolating these pixels in the correct positions. For example, when converting from 625/50 to 525/60, 714 active pixels from a 625 line will be used for the computation of the 702 pixels of a 525 line.

The change of the number of lines (the vertical conversion) is more complex. It is not only necessary to deal with a considerably larger amount of data to be processed in order to achieve a good interpolation, but it is also necessary to take into account that all present-day standards are interlaced. Consequently the interpolation cannot involve all of the available data inside one field and change the number of lines in that field since they represent only one half of the total vertical information. Therefore, to achieve an excellent vertical conversion, it is necessary first to use sophisticated methods to deinterlace the frame and then to compute the lines on the basis of data from three consecutive fields. Such conversion does not require a large amount of memory and can be made virtually transparent.

The conversion of picture rates is the most delicate problem. Not only are the number of fields different, but this conversion is also correlated to the sampling of movement. In a 625/50 standard, each movement is sampled at the rate of 50 Hz, that is, decomposed into 50 static pictures each second, while in the 525/60 standard, the same movement is decomposed into 60 static pictures. Therefore, when converting from one standard to the other, it is necessary to take care not only of the number of fields, as such, but also of the preservation of the quality and smoothness of the movement rendition. For that reason this part of the conversion process is called temporal conversion since it consists of converting from one temporal movement representation to another.

The simplest and the crudest way to make such a conversion would be to simply drop every sixth field when converting from 60 to 50 fields or to repeat every fifth field when converting in the opposite direction. However, the lack of one phase or the repetition of one phase of the decomposed movement would create a defect; either would appear as repeated jumps in movement rendition. A better solution would be to generate a new field that has to be added in the 50 to 60 conversion on the basis of the data collected from several fields. This method eliminates the jumps but still does not insure a smooth movement rendering. In the case of a moving object in a shot or of a camera movement, the data collected from several fields will also register the object or the background in different positions. The resulting picture will display a lack of movement smoothness, something like a hesitation or a trembling of the edges, called judder.