A Hollywood (CD) player is where it’s at

The following article is based on a presentation by Andy Bechtolsheim, co-founder of and vice president of technology for Sun Microsystems, at the Digital World conference in June. There, he detailed his concept of a digital-movie player based on technologies that exist today. Bechtolsheim opened by saying, “Sun will not build this, but if I were Sony or Matsushita, I’d be crazy not to.” Considering his track record, it would have been pretty dumb not to listen. The audience responded in kind — nearly every hand in the house shot up when he asked if they would buy such a player for $1,000.

We are here [at Digital World] because of the success of digital compact discs and digital memories, not because we have been waiting for digital transmission technologies.

Perhaps the single most important opportunity in digital media at this time is to set a standard for high-quality digital video and to deliver mass-market products based on this standard. Similar to the great improvement in sound quality achieved by digital CD-audio, there is clearly a major consumer demand to view high-quality movies in the home.

The player I’m about to describe could be available in two years. While a digital video player does not reduce the future potential of digital television broadcasting and cable, it is not dependent on digital broadcasting either. Thus it will deliver the benefits of digital, high-quality movies without changing the infrastructure required to deliver them.

There are a number of bottlenecks to widespread acceptance of digital television broadcasting. One is the regulatory bottleneck of the U. S. Federal Communications Commission to select a standard. Another is a capital bottleneck: broadcasters will upgrade to expensive digital equipment slowly, only when there is sufficient consumer demand for digital TV. And on the telecommunication side, there is a transmission bottleneck because our homes are wired with copper cables instead of high-bandwidth, fiber-optic cable.

Next: Digital video before digital TV
Because of these bottlenecks, digital-media video will achieve market penetration a long time before digital television. The question is, “How long?” My estimate is that it could take five years until digital television will achieve the same installed base as digital media video systems.

Digital media technology is here today. It costs only $1 to produce a compact disc. The distribution channels are well established. Retail outlets such as Tower Records will sell the video CDs, the Blockbuster Videos of the world will rent them, and stores such as Circuit City will do a brisk business in the sale of digital video disc players.

It’s obvious that standards are good, but in the end, great products are better. A high-quality digital video player could be launched by whoever has a dominating market share to set a media standard. This requires strength in both movies and consumer electronics. Two companies, Sony and Matsushita, already have both in place. Now we need the product — a high-resolution VCR replacement to display high-quality, digital movies.

The player’s most important purpose is to play digital movies. If it can also be a video game machine, that’s even better. But it’s not a computer. It must be a mass product from the start, since mass products are what set standards. It also must cost $1,000 or less.

What will make this product possible is the next generation of CD technology, already in the labs, which will have the storage capacity required for high-quality digital video. The level of resolution, which needs to be defined, will be somewhere between today’s NTSC standards and the HDTV format. Of course, the exact format depends on the storage capacity of the media, the video compression achieved, and the display resolution and sample rate. In my formula for the player, I am assuming a display resolution of 9605540-pixel non-progressive scan at 24 frames per second.

The perceived quality of such a display will be about one-half of interlaced HDTV and four times better than NTSC. This level of quality seems to be a good choice. Whatever the final standard is, it must be set with care, since media standards generally last 20 years or more.

Next: What are the applications?
Besides playing movies, other applications are likely to include games, books and interactive programming; these will require programming interfaces, additional data definitions and so forth. But what’s more important is to set the standard for high-quality digital movies on compact disc.

All of today’s CD-based products are red-laser-based. To establish a new media definition, it’s critical to use the best technology that you can get in quantity. A green-light laser technology, which I call cd2, pro-vides twice the density of today’s CDs, and is already close to market. The best technology available in the labs is blue-light laser, cd4, which allows four times as much data on a CD as the present red-light laser technology.

APPLICATION REQUIREMENTS
The media specifications for a digital video player are determined by the application requirements for inter-active tasks such as audio, video, movies, electronic books, images (such as an image of a page) and 35mm photos.

For CD-quality audio, sound must be sampled at 44 khz. Compressed at 5:1, it requires 30 kb per second of playback, which achieves very good sound quality as demonstrated by the new Sony magneto-optic Mini Disc format that uses this level of compression. To get surround sound will require an extra channel, for a total of about 50 kb per second.

For NTSC-like picture quality, let me assume a resolution of 4805270 pixels and the ability to store the data at 24 frames per second, the rate at which most movie material is recorded originally. With 100:1 mpeg compression, to play one second of this type of video requires 62 kb of storage.

Clearly we must have much higher quality than NTSC in this product. Four times the resolution of NTSC requires 9605540-pixel resolution. Using the same data rate of 24 hz and 100:1 compression, to play one second of digitized film stock requires 250 kb of storage.

(Note: MPEG isn’t yet quite good enough to do high-quality compression at a 100:1 ratio. This seems like an achievable goal but will require more work. Also, audio isn’t included in these figures — that needs to be added on top. For movie and surround sound audio combined, the player would need to be able to have a data rate of 300 KB per second.)

Next: Electronic books, images, photos.
If the digital movie player can handle 300-KB-per-second playback, then 30 kb can be displayed per “page” (read: screen) at a rate of 10 hz, which is an acceptable page-turning time. This is plenty for textual information, since a single text page requires about 10 KB of information and even less with compression. A full-page image compressed 20:1 with JPEG requires about 30 KB, which still satisfies the 10-hz page-turning goal. Photographic-quality images, such as the Kodak Photo CD provides, require much more storage — about 6 MB per photograph. With 20:1 JPEG compression, the image requires 300 KB of storage. This allows users to view photographs at one picture per second, although subsampled photographs could be viewed at movie rates.

WHY PUT MOVIES ON CD?
The advantages of mastering movies on CD become obvious when compared to other media. A floppy disk, though it costs the same as a CD (about $1), is too small for anything. Sixty minutes of audio on a floppy would cost $100; 120 minutes of a digitized film-quality movie, $1,000.

One movie, one disc. Magneto-optical, rewritable discs are much more expensive — $10 each — and though densities are increased, it would still cost $160 to master a 120-minute movie onto an MO disc. Even if a single MO disc could hold a full movie, studios would not be anxious to use a writable medium — it would make consumer copying much too easy. Read-only media is the preferred media by the software industry. In addition, MO discs are not compatible with the audio CD format.

Compact discs are the most cost-effective digital media today, with a manufacturing cost of about $1 per disc. The main limitation of today’s CD format for movies is the limited capacity. Assuming a 300-kb-per-second data rate for high-quality digital movies, an audio CD would hold only 30 minutes of movie material. Four or five discs would be required to hold a feature-length movie. CD2, the double-density green-light laser format, still doesn’t achieve the goal of one movie-one disc. Only the blue-light CD4 has sufficient data density to fit an entire movie on a single disc.

Next: COMPRESSION AND DISPLAY ISSUES
The MPEG standard, still being refined, will include several different resolution and sampling rates. The trick for this product is to make the right choice about the key parameters that provide the best display quality to the consumer while meeting the constraints of the available media storage capacity and data rate.

Square pixels are absolutely essential for optimum data storage, compression and compatibility with future computer applications.

As mentioned earlier, there is a big advantage to storing video on digital media at 24 hz noninterlaced (progressive scan), since this is how virtually all movies are originally recorded. Any digital system requires a frame buffer for decompression, so the obvious approach is to refresh each picture three times at a 72-hz screen refresh rate, which also solves the flicker problem that is present in the NTSC format. In comparison, a 60-hz interlaced NTSC approach wastes bandwidth, makes compression more difficult, reduces picture quality and causes refresh flicker.

Players must offer scalable resolution to support both existing NTSC/PAL-format displays and the new digital video format. In addition, to make such a system forward-upgradeable, attention must be paid to upcoming digital video transmission standards.

THE PRODUCT LINEUP
The following product lineup reflects the above requirements.

* 28-inch digital video display — $1,195
Noninterlaced, 28-inch CRT, with 16:9 aspect ratio
960×540 resolution with square pixels
Digital video input from CD4 player
Analog TV receiver for NTSC/PAL viewing with “letterbox” capability for NTSC/PAL

* Stationary player for home use — $595
CD4 video/audio player
Optical digital connection to digital video display

* Portable digital video player — $1,195
Noninterlaced, 6-inch color TFT LCD display with 16:9 aspect ratio
480×270 resolution for subsampled video
CD4 video/audio player in same enclosure
Optical digital connection to digital video display

THEY HAVE THE TECHNOLOGY
The technology exists now for a low-cost, consumer digital video player. Though Sun is not planning such a product, several Japanese companies have the technology to make such a product, independent of standards, regulatory bodies or broadcast issues. I expect to see this kind of product on the market within two years.

- Andreas Bechtolsheim, with Denise Caruso