Archives: July 2007

15/07 Red Bull get it wrong

If you go to a cinema in Greater London this fortnight you have a good chance of being subjected to "the world's first 3D cinema advert", advertising some event being run by a well-known brand of fizzy pop. I'm not sure if this was gimmickry for the sake of it, or if Red Bull want to be part of the current wave of 3D cinema, but I'm sorry to say that I wasn't impressed.

First of all, to avoid the need for digital projection equipment or dual-projector rigs for a 30 second advert, Red Bull chose to use the anaglyph method of generating a stereoscopic image, and corresponding red- and cyan-lensed "3D glasses" were handed out before the show. This technique is decidedly non-ideal for cinematic presentation - the fact that the colour information is split between the two eyes produces a highly unrealistic effect.

The inherent problems of the anaglyph were compounded by a botched process in the mastering and printing of the film - the red and blue/green light that was projected was a poor match for the filters in the 3D glasses, with the effect that there was considerable cross-talk between the two eyes, making it virtually impossible to converge the two images and experience a sensation of 3D viewing. To make things worse, the (mostly computer-generated) material made use of considerable parallax for effect, and consisted of a series of short, fast-moving scenes - a typical advert in that respect, but absolutely the wrong way to go about giving the audience a good experience.

If I was in the business of trying to promote three-dimensional cinema, I'd be quite annoyed, I suspect...

10/07 Kingswood Warren in the news

It seems that the journalists from CNet Crave have been visiting my department to have a look at my colleagues' prototype system for doubling the capacity of a DVB-T multiplex (the technology used by Freeview) to try and make terrestrial HDTV practical. The technology makes use of the fact that you can transmit two independent signals on the same frequency if you polarise them 90° apart - something that's been known for about a century, but hasn't been practical until recently, with the development of MIMO techniques.

They are currently squeezing three HD transmissions and one Standard Definition TV channel into a single multiplex, the HD channels at a bitrate of 15Mb/s - plenty for the most recent version of Dirac (although as far as I know nobody's proposing that codec for transmission) and h.264 codecs ought to be looking pretty good at that bitrate by the time any hypothetical MIMO technology could be deployed country-wide. (The current BBC HD trial service started broadcasting using h.264 at 19Mb/s back in 2006, and at that time it really needed a bit more than that, in my opinion. It is in the nature of video codec implementations to improve with time, however, as their creators find ways to make better use of the data-reducing tools that the codec standards provide.)

No technology is ideal, and this one would require everyone to upgrade their aerials again, as well as their set-top boxes, but to get HD into the very limited bandwidth available terrestrially requires compromises, and this technology is a very interesting option.
  • Time: 04:47PM
  • Category: BBC

06/07 A little more evidence

In a recent post, I was talking about why televisions refresh their picture at the same rate as the mains frequency. As far back as 1939, I said, flicker caused by the mains-driven studio lights beating against the picture-taking frequency of the studio cameras was being cited as a reason. Supporting (if anecdotal) evidence comes from "An unreliable and wholly unofficial history of BBC Television Centre", by Martin Kempton, which recounts how studios TC6, 7 & 8 were equipped in the 1960s with kit to allow them to create programmes in the American NTSC standard. This involved running the cameras (and other equipment) at a field rate of 59.94Hz, whereas the studio lights were still driven at the UK mains frequency of 50Hz. Martin reports that flicker caused by beating of this kind was indeed a problem when the studios were operated in "NTSC mode". It seems that few NTSC programmes were made at TVC, in part because of operational issues such as this one, but also because high-quality "standards converters" for turning PAL programmes into NTSC ones were entering operational use at around the same time as these studios were commissioned (including the "field-store" converter developed here at Research Department).

As we investigate the effects of higher frame rates on video quality, this is something we will have to bear in mind...
  • Time: 01:21PM
  • Category: BBC