No. 301 (January 2005)
The EBU Technical Committee has recently issued two recommendations related to HDTV: one concerns emission standards, whilst the other concerns production standards.
On emission standards, the EBU favours the use of progressively-scanned formats, such as 720p/50 or 1080p/50, rather than interlaced formats such as 1080i/25.
(N.B. The convention used to describe TV formats is the “number of active lines per frame” + the "scanning algorithm” [interlace(i) or progressive (p)] / the “frame rate”. Current and near-future scanning formats include 576i/25, 720p/50, 1080i/25, 1080p/25 and 1080p/50.)
On production standards, the EBU recognises that there may need to be a wider range of formats and suggests that broadcasters should buy equipment that supports, at least, three formats: 720p/50, 1080i/25 and 1080p/25. Although 1080p/50 is not yet a viable option, it is an attractive production format for high-quality production – from which 1080i/25 and 720p/50 versions can easily be derived. As 1080p/50 offers future-proofing of valuable content, the EBU is keen to promote its adoption in the longer term.
Why is the EBU so firmly in favour of progressive scanning? Before answering this question, it may be helpful to examine the principles of interlaced scanning and progressive scanning.
 |
 |
| Figure 1: Interlaced scanning | Figure 2: Progressive scanning |
Although both of the above animations show that the scanning of the entire picture takes the same duration with the two scanning processes, interlaced scanning is an old but effective technique for bandwidth reduction. Engineers discovered in the early 1930s that the progressive scanning process needed to scan the picture about 50 times per second whilst the interlaced scanning process could operate at half of that speed.
Modern compression systems offer much better performance than the fixed technique of interlacing: they combine several techniques which can be optimised to suit the picture content. All compression schemes deliver better results at lower bitrates with progressive sources. As modern HDTV displays are inherently progressive, it is necessary for these displays to include converters from interlace to progressive. Hence, progressive scanning is the logical choice for emission. Furthermore, if sources (i.e. cameras) and displays are both progressive, use of interlace should be avoided anywhere in the signal path.
Format “wars” in the USA resulted in the ATSC standard
including 18 scanning formats. The heated debates about formats at IBC 2004
caused many people to think that Europe’s renewed interest in HDTV could
be damaged by further format wars.
The agreement on the common image format of 1920 x 1080 pixels for HDTV was
a remarkable achievement, but various parties (including the EBU) are now advocating
the use of 720p/50. What is the difference between the 720p/50 and 1080i/25
formats?
The advocates of 1080i/25 will tell you that 1080-line systems offers 2.07 million pixels (= 1920 x 1080) whereas 720-line systems offers only 0.92 million pixels (= 1280 x 720) – corresponding to a ratio of 2.25 between 1080 and 720 line systems. However, we must avoid comparing “apples” and “oranges”. If you look carefully at interlaced pictures, you will see “inter-line twitter” which reduces the subjective vertical resolution: many tests indicate that, in subjective terms, 1080i and 720p deliver the SAME vertical resolution.
Nevertheless, 1920 x 1080 systems potentially give greater horizontal resolution than 1280 x 720 systems. In fact, the ratio of 1280/1920 indicates that 720 line pictures can deliver 66.7% of horizontal resolution. In practice, about 95% of current 1080i cameras have 1440 pixels horizontally (rather than 1920 pixels), thus reducing the ratio to 1280/1440 = 88.9% of horizontal resolution. We are told that in the future all 1080i HDTV cameras will offer the full 1920 pixels – but many popular models are limited today to 1440 pixels. Similarly, few displays can today reproduce the full horizontal resolution of 1920 pixels – but this may be a temporary problem until 1920 x 1080 displays become commonplace.
In summary: 1080i has the same vertical resolution as 720p, whilst 1080i has slightly better horizontal resolution.
Static resolution is important for still pictures, but very few TV pictures are “static” e.g. most sports coverage involves rapid movement and unpredictable camera zooming. Everybody seems to agree that progressive scanning gives improved motion portrayal – especially for slow-motion. The choice between 1080i and 720p is thus a balance between static resolution and motion portrayal: 1080i offers better static resolution whereas 720p offers better motion portrayal. Which is more important to you?
 |
 |
| Figure 3: Interlaced scanning (©SVT) | Figure 4: Progressive scanning (©SVT) |
It is important to recognise that it is easy to convert a progressive format (e.g. 1080p/50) into an interlaced format (such as 1080i/25), but it is much more difficult to convert an interlaced format into a progressive format. Fig. 3 shows an interlaced image whilst Fig. 4 shows a progressively-scanned image. As the camera was being panned to track the motion of the jogger, the effects of interlacing are most obvious on lines which are vertical or nearly vertical, such as the trees. In practice, most of the “cog-shaped" appearance of the trees in Fig. 3 are smoothed out by the temporal characteristics of the displays, the human eye and the brain. Even so, if you can imagine that you were a video compression system, my guess is that you would prefer to process Fig. 4, rather than to try to remove the interlacing artefacts from Fig. 3.
Of course, you could argue that camera panning is predictable and therefore you can easily remove the worst of the interlace artefacts – but those who have worked deeply on video compression schemes will tell you that simple problems have a habit of becoming very complicated. Real pictures contain objects (e.g. people) who move unpredictably or suffer from the problem of “revealed motion” in which moving objects can be temporarily obscured by other stationary or moving objects in the picture.
At an IBC 2004 session, I was asked to explain why format wars seemed to be on the horizon for Europe. Having explained the EBU’s position in favour of progressive formats, one member of the audience said:
“I am amazed that anybody would consider launching new services based on interlace. I have spent all of my life working on conversion from interlace to progressive. Now that I have sold my successful company, I can tell you the truth: interlace to progressive does not work!”.
This remark was followed by a spontaneous round of applause from the audience. When the person was asked to give his name, he replied “Yves Faroudja” (who is widely acknowledged as the world’s foremost expert on format converters).
At this point, I should declare a personal interest in Faroudja’s opinion: when I was purchasing a 50-inch display one month before IBC, I was very aware of the dismal performance of some format converters. Why did I buy this particular display? Mainly because it was equipped with “Faroudja” filters …..
It is worth examining the longer term role of 1080p/50 and 1080p/60, which represent 3rd generation HDTV. For these to be adopted as production standards, studio interfaces must be defined. As 1080p/60 has an uncompressed data rate of about 3 Gbit/s, this is not a trivial requirement. However, if we could achieve lossless compression – reducing the data rate by a factor of 2 – the infrastructure designed for 1080i/30 could be reused without problems. Despite such challenges, 1080p/60 and 1080p/50 will become attractive production formats for high-quality production: for example, the full resolution offered by these formats ensures that valuable content can be future-proofed. As it will be some years before 1080p/60 and 1080p/50 can be considered as emission formats, we may by then have benefited from further advances in compression systems. Even so, interfaces between set-top boxes and displays present several problems.
Some suppliers of HDTV equipment have complained that the EBU’s support for progressive scanning is damaging the case for 1080i/25 and for the 1920 x 1080 common image format. In fact, the EBU has recognised that 1080i/25 services will operate alongside 720p/50 services – and strongly hopes that 1080p/50 systems will eventually become the norm.
The EBU has no intention of starting a format war or prohibiting the use of 1080i/25. It merely wants to ensure that broadcasters will have the freedom to select the format that suits their needs and the needs of their audiences. Fortunately, as this editorial was being written, the members of EICTA (the trade body representing the consumer electronics industry in Europe) have just agreed an HD-ready label to help consumers understand whether a particular display is ready for HDTV. In essence, this HD-ready label can only be attached to displays that meet the following conditions:
This breakthrough is very important because it demonstrates that the emergence of two competing formats (720p/50 and 1080i/25) for HDTV in Europe will NOT cause problems for consumers. It also shows that there is no need for a format war: the two systems can coexist peacefully with each other.
Commercial and political pressures have long influenced the selection of TV standards. In setting HDTV standards, we must avoid “half-truths” about technical matters. The EBU plans to undertake independent subjective tests that will evaluate many of the assertions made by the advocates of the competing standards. The EBU wants broadcasters to be able to make “educated choices”.
Philip Laven
Director
EBU Technical Department
26 January 2005
| European Broadcasting Union L'Ancienne Route 17A CH-1218 Grand-Saconnex Geneva Switzerland techreview@ebu.ch |
 |