Instead of just two basic standards, HD runs to 26, and there seem to be additional standards appearing every month. There are different meanings these days to the words ※standards converter§. What about audio? Not to mention all the other data that makes up content today.
So let＊s try to clear away the confusion and provide answers to that difficult question: what is a standards converter? Firstly, let＊s define what standards are covered - what signals can go into a standards converter, and what signals can come out.
All television signal standards divide into one of two basic groups, generally depending on the electrical power supply frequency of the area where the signal is produced. For most of the Americas this is 60Hz, while for Europe it is 50Hz. In practice professional TV signals in North America are based on a 59.97Hz value.
What＊s the difference? Simplistically, there are 20 per cent more pictures per second with 60Hz than with 50Hz. If you are converting from 60Hz to 50Hz you could just throw some of the pictures away 每 and hope no one will notice. Or going the other way, you could duplicate the occasional picture, but that might produce some even more obvious glitches. Back at the issue of standards, the two basic groups divide up into a whole series of sub-groups.
Standards conversion relies on the process of interpolating picture content 每 either because you need extra pictures (going from 50 to 60Hz) or because you need less. In both cases it＊s necessary to create images that simply did not exist in the original picture stream. There are several ways of doing this.
If you don＊t care how long it takes to do it, you could feed the original pictures into a conventional computer, and provided it has enough of them, it could work out what the ※intermediate§ pictures should contain, and create them accordingly. The technology to do this exists, and it is an option if the requirement is simply to convert content files between formats. But it＊s of no use whatsoever if the content is ※live§, or even if the turnaround time needs to be measured in minutes rather than hours. So it＊s simply not an option for content like sports or news.
A crude method that would work with live material is to store several pictures, and repeat the odd one now and again to get up from 50 to 60. The trouble with this is twofold: on the one hand you would see a glitch every time there was a repeat picture; and on the other, the actual time covered by the ※non-repeated§ pictures would be different to the original time covered.
The best way to convert is to create a completely new stream of pictures, based on the originals, but completely interpolated to the correct information for the required output standard. The simplest way to do this is to store several and then average across them to produce a new stream. This is fine for still pictures, but would just produce a blurred mess if they had movement. In fact there are many low-cost standards converters for SD that use this method 每 normally using four fields of storage. The results are just about acceptable if there＊s not too much rapid movement 每 but the technique is simply not adequate for HD.
What is required is a way of estimating where elements of the interpolated picture would be, based on the direction and speed of movement of those elements. The direction and speed values can be expressed as vectors, and the mathematical values of these vectors can be used to ※move§ the elements to the correct position in the new interpolated frame.
Let＊s use a very simple example 每 a vertical black bar moving from left to right across the picture. For simplicity, this bar is only 1 pixel wide. As it moves from one picture moment to the next, it＊s fairly easy to see that each pixel can establish a direction and speed of movement 每 its vector value. Knowing this, it＊s possible to calculate the exact position of each pixel in the line at any moment in time 每 including at a completely different picture rate.
This motion estimation process is very complex, and requires considerable computing power, not to mention sophisticated algorithms, to produce a satisfactory result. But when applied correctly the resulting pictures cannot be distinguished from the originals 每 except that they are at a different picture rate.
The impact of motion estimation is immediately visible on any picture sequence involving horizontal motion. Take the example of a typical news or sports presenter situation with a fairly static picture of the presenter and a tickertape running across the bottom edge of picture. Look at that without motion estimation and you will not be able to read any of the tickertape text. With motion compensation it＊s indistinguishable from the original.
There are other key issues in the HD world. Anyone with experience of standards conversion will tell you that a standards converter is a standalone box, complete in itself. That was certainly true of standards converters up to now. But today＊s television systems are much more complex than they used to be, and a single-use box is no longer acceptable.
A prime example is that of audio. Back in the early 90＊s all that a standards converter had to do was convert video pictures. Any audio was routed around the converter. Today, it＊s much more likely that audio signals will be embedded in the video stream, possibly along with a range of other metadata. So it makes much more sense to design a modular HD standards converter, where different modules can be added to the basic video converter to support different demands. You might just need to convert video 每 if so you＊ll only need the converter and motion estimator modules. But if you need to convert other parts of the digital signal stream, you can simply add modules to suit.
Pro-Bel＊s Vistek product division and Digital Vision have probably more experience of motion estimated standards conversion than anyone else. Digital Vision＊s considerable expertise is utilised to calculate the motion vectors, while Pro-Bel＊s deep understanding of HD conversion processes produces the resulting converted picture stream. Between the two organisations there＊s over 60 man-years of standards conversion know-how 每 indeed the most senior members of the development team worked on the world＊s first digital standards converter back in the late 70＊s.
But unlike those early converters, CIFER is truly universal because it can convert between any SD standard, and between the primary HD standards, and from SD to HD or HD to SD. It simply doesn＊t care 每 throw any standard at the input, and it will convert to the required output standards 每 without any loss of picture quality.
CIFER will convert between all of them, regardless of frequency. Only a true standards converter can do this 每 don＊t be fooled by boxes that can convert between different line standards 每 720p60 to 1080i60, for example 每 because they can＊t convert between 50Hz and 60Hz frame rates. It is also modular to allow the addition of a huge range of signal conversion solutions.
Let＊s take embedded audio again as the example. For simple applications, CIFER can be specified with internal embedded audio processing, which will ensure that the input embedded audio is re-inserted with the correct delay at the output stage. Then there＊s an Advanced Audio Processing module (V6302) that can strip the audio from the video, delay it to compensate for the video delay through the standards converter (and swap channels or adjust levels if this is required), and then re-insert it to the correct standard for the output picture stream. If the audio is in Dolby E form, it＊s necessary to de-code it and then re-encode it to the correct picture rate (Dolby E needs to be locked to picture rate). In this case two audio processing modules are then required. These are the V6302-DB and V6302-EN. They incorporate the relevant sub-module options for Dolby E decoding and encoding respectively.
We live in a multi-standard, multi-format world where signal conversion is of paramount importance, alongside flexibility and the ability to respond to change quickly and cost-effectively. HD demands absolute quality and CIFER delivers.
|
