The importance of a scaler

The aim of a home cinema installation is, by means of a good multichannel sound system and very high-quality images, to bring into our homes the same sensory experiences we can have in a commercial movie theater. If we do it well, we will even surpass the most modern cinema.

Cinema (commercial and home) is nothing more than an attempt to represent reality as faithfully as possible, to convince viewers that what they are seeing "exists" (even when the genre is fantasy). To do this, the image must be good and believable enough to trick our brain into thinking that fiction is reality. One way to achieve this goal is to break through the barrier of the television set, with an image large enough to fill our field of vision. This is where the video projector comes into play.

But out of technical necessity, and given how slowly any standardized initiative aimed at delivering an image with suitable quality and resolution for this purpose gets off the ground, we need to use some form of video processing that allows the old PAL broadcasting standard (the basis for the vast majority of modern standardized video formats, such as the one used by DVD-Video) to be projected onto a 100" screen without the defects and artifacts inherent to a solution designed for a 4:3 aspect ratio and screen sizes of barely 29". Welcome to the video processor and scaler.

The PAL standard has a resolution of 720x576 pixels (NTSC even lower: 704x480). This means that an image on the television screen is made up of 414,720 dots. Digital television (terrestrial and satellite) only changes the method of transporting that same standard. DVD-Video, with its MPEG, does the same for the storage system. The old PAL standard, designed for a specific situation and specific televisions (4:3 and up to 29" diagonal), is and will remain present in our lives for some time.

But we are increasing the size of our displays (from 42 to 100 inches or more), making the limitations of the current PAL standard obvious. Until high definition (HDTV) arrives, our concern is to be able to continue enjoying our DVD-Videos and, why not, television on these larger screens.



The deinterlacer



Both PAL and NTSC use a scanning technique known as interlacing. This is a compression system that takes advantage of the sluggishness of the human eye and the technical characteristics of the phosphors in cathode ray tube (CRT) televisions. The advantage is that it reduces the bandwidth needed during transmission (especially terrestrial transmission), although the disadvantages (more evident the larger the screen, especially beyond 4:3 and 29") are flicker and motion artifacts.

Interlacing allows a reduction of up to 50% in the information to be transmitted per unit of time. The human eye has two important characteristics: sluggishness and resolution. The first means that if 24 still images are displayed per second, we are unable to perceive them as still, but rather as moving images (which is why cinema can project 24 "slides" every second and we see "motion"). The resolution of our eye is 1º; in other words, we are unable to make out anything below that angle.

When the beam of a cathode ray tube strikes (and therefore excites) the phosphor of a cell, the cell lights up, converting the energy into light. But it does not go out immediately; it takes a certain amount of time to do so.

If we combine these three characteristics (resolution and sluggishness of the human eye and phosphor persistence), we will understand the reason for interlacing. A still image (of the 25 that make up one second in PAL) in broadcasting is called a "frame". Each frame is divided into two fields. The first field contains only the visual information of the odd lines, while the second contains the even lines. A tube television reproduces an image from top to bottom and from left to right. But instead of doing so line by line, it first scans the field of odd lines and then that of the even lines. When this second field is scanned, the phosphors of the previous field (that is, of the odd lines adjacent to those now receiving energy) are still lit. Our sluggish eye perceives the two fields as if they were a single image. Each scan (odd and even) is carried out 50 times per second, which is why we say that PAL has a frequency of 50 Hz, even though only 25 still images are broadcast.

But as we increase the size of the television, we will gradually notice the defects inherent to interlacing. First we will notice how an annoying lack of definition, a stair-stepping along the outline of the moving object, appears in fast movements. If the whole camera is moving, the blurring will be general. The second, no less important, defect is flicker. As the screen size increases (but the resolution of the human eye remains the same), not only will we see dots instead of images, but we will more easily perceive each pair of lines switching on and off.

Deinterlacing the image becomes imperative to solve some of these problems, all the more so since, with the exception of traditional CRTs, all modern displays are not made with phosphor cells but with other materials with very different physical and chemical characteristics.

A deinterlacer is a complex digital circuit whose results vary depending on its design. Most use complex mathematical algorithms to combine the fields.

The line-doubler type of deinterlacer is the most basic. Its main purpose is to merge into one the two interlaced fields that make up a frame. When this is not possible (for example, in scenes with a lot of motion), it simply doubles the lines. In other words, in an odd field, where the information is on lines 1, 3, 5, etc., it copies the information onto the even lines: line 1, line 2 with the information from line 1, line 3, line 4 with the information from line 3, and so on. This solution does deinterlace a signal, but it causes 50% of the vertical resolution to be lost. Losing this information when we are increasing the screen size is undoubtedly a step backward.

An advanced solution is to use interpolation techniques. This makes it possible, to a greater or lesser extent depending on the complexity of the circuit used, to predict the best information between two known points. Instead of "copying" the information from the known line above or below, an "average" of the two is calculated, so that the missing information (that is, the even lines in odd frames or vice versa) is as realistic as possible. This method gives the viewer the illusion of higher resolution. Obviously only 50% of the vertical resolution is retained, but depending on the type of interpolation used, the result comes more or less close to reality.

The next step up in quality and complexity is motion-adaptive processing. While the previous method can maintain and even exceed the perception of higher resolution, motion artifacts are still evident in fast movements or camera pans. Motion-adaptive deinterlacers use even more complex circuits, since they try to identify moving objects, predict their direction and speed and attempt a much smoother transition. During processing, these solutions take into account not only the information from adjacent lines, but also the nearest frames and fields. The most advanced are now called "motion-compensated" deinterlacers, which include complex algorithms even to recreate parts of objects that can barely be made out in the original image (that is, the interlaced video signal).

We have to be aware of and acknowledge the fact that no solution can deliver a very high-quality image where no information exists. Moreover, there are random movements (such as the waves of a sulking ocean) that are almost unpredictable even for the best mathematical algorithm.



The film advantage



One of the advantages of the PAL format is its frequency of 50 Hz which, as we have seen, actually means 25 images per second. Cinema uses a rate of 24 frames per second. During the telecine process (that is, from film to television), the film is played at 25 frames per second, matching each film frame to a video frame. The differences between fields within the same frame are then nil, which means that even a mediocre deinterlacer achieves good results. In fact, motion artifacts are lower than with video-originated programs. Things are more complicated when we have an NTSC video made from a film, which, because of the higher frequency of this television standard (60 Hz), uses a different technique.

If, during deinterlacing, the system is able to recognize the film source (or lets us select this option), the results on screen can be much better.

The MPEG on a DVD-Video disc allows a flag to be included to identify its film origin, making the deinterlacing circuit's task easier.



New resolution



Not all modern displays have a native resolution of 720x576 pixels. First, because we are waiting for HDTV, which will allow us to increase vertical and horizontal resolution natively (anyone who has enjoyed a high-definition image will want nothing more to do with standard PAL resolution). This bet on the future (which includes the 16:9 aspect ratio) is what justifies the current investment in high-resolution displays. But until high-definition images reach us in a uniform, standardized way, some kind of processing capable of increasing and changing the resolution of the PAL standard is imperative.

As with deinterlacing (the step prior to resolution conversion), there are different ways of achieving this goal. From straightforward line duplication to the various interpolation methods, that is, once again the mathematical recreation of data that does not exist.



The scaler



All modern displays have their own deinterlacer and scaler. Some are proprietary solutions, such as Philips' Pixel Plus 2, others are commonly used ones, such as the Faroudja/Genesis deinterlacers, and others use OEM circuitry.

External scalers (that is, devices placed between source and display to carry out deinterlacing and scaling) are now becoming popular thanks to the new digital links. Until now, the necessary D/A and A/D conversions degraded, albeit minimally, a signal that we wanted to improve.

Thanks to digital connections, these conversions disappear. Previously, external scalers could deliver their video output through an analog PC connection. An advantage if the display was compatible.

In video projectors, where we can easily aim for screen sizes of 100 inches or more, the video processing circuit (including deinterlacing and scaling of the image) is as important as the quality of the panel and the optics. In order to keep production costs, and therefore the final retail price, low, many manufacturers use the bare minimum of electronics needed to display images correctly.

Using an external scaler does not mean working magic, far from it, but it does mean significantly improving on the quality of the circuitry that a particular projector may have. In addition, the advantage of a good scaler is that it can feed several displays: from our projector to the 40" high-resolution plasma TV.

Some scalers can even be updated on demand to new resolutions through software updates. It is all a question of weighing up the investment a scaler represents against the improvement in the picture.