Then and Now

by Michael Karagosian
©2005 Karagosian MacCalla Partners, all rights reserved worldwide
Published in the April 2005 issue of INS Asia Magazine

Star Wars has been the poster child for digital cinema since Episode I was released in 1999. It was with the first motion picture release of Episode I that four digital cinema projection systems were installed in the United States: two in Los Angeles, and two in New York. The motion picture industry has not been the same since. Many companies since attempted (and failed) to roll out digital equipment for motion picture exhibition, the major movie studios banded together to agree on final formats for digital distribution, and exhibitors around the world continue to grapple with how to afford what will be the most expensive equipment upgrade in the history of motion pictures.

Last November I sat down with Rick McCallum, famous as George Lucas' producer for Star Wars Episodes I, II, and III, to talk about our favorite subject. With resignation, Rick informed me that Episode III, to be released this May, would be seen in two less digital screens in the United States than Episode II. This is disappointing news for George Lucas, probably the most renowned pioneer in the introduction of digital technology to the motion picture industry. However, each new release of the Star Wars trilogy has been a milestone for progress in the industry. The upcoming release of Episode III will be no different. The short history of digital cinema deserves a review.

I already mentioned that for the release of Episode I, two digital cinema projection systems were installed in Los Angeles and two in New York. In each city, one Hughes/JVC ILA-12K projector and one prototype DLP-Cinema projector were deployed. For several years prior, the Hughes/JVC projector had been the cornerstone of digital cinema conversations. It was Episode I, however, that demonstrated the more robust DLP Cinema technology as the superior technology. This was a significant turning point, as it was clear that there was now a digital projector that could withstand, without sensitive realignment, the 14 hour day, 7 days a week workout that commercial projectors would have to endure. While there was still room for improvement in the DLP technology, the low 1280 x 1024 resolution being one, the technical problem for digital cinema was no longer the projector, and the focus shifted to the server.

Episode I deployed Pluto Technology servers, storing a lightly compressed D5-format digital copy of the movie on hard drives, allowing it to play the movie on demand in the theatre. Following the release of Episode I, experimentation in digital cinema continued with the combination DLP Cinema projector and QuVis 'QuBit' server. The QuVis server utilized a proprietary wavelet compression, which delivered a very high quality image using much less hard drive space that the original Pluto servers. While it was still awkward to distribute the necessary 30-50 GBytes of digital movie to each server, it was far easier to manage this than to deliver 1 TByte of data to the Pluto. Later tests in the area of compression demonstrated that regardless of compression technology chosen, the amount of compression required for a true visually lossless condition was about the same. This was a revelation, and the search for compression technology shifted from proprietary solutions to openly (and inexpensively) licensed technologies. Thus, MPEG2 became a viable compression technology for digital cinema.

In 2002, Episode II was released to approximately 50 digital screens in the United States. The projectors were based, of course, on DLP Cinema technology. The servers, however, were of several types. The original QuBit servers were still in use, but for the release of Episode II, new and different servers were installed, some based on Qualcomm's ABSOLUTE compression, and some based on MPEG2. All-in-all, at least four different digital masters of the movie had to be created to allow digital playback on the available systems. Over the years following Episode II, the industry attempted to reduce the number of different types of digital distributions required.

Episode III will be released in May to a somewhat improved digital cinema scenario. While many of the projectors in use remain the older 1280 x 1024 resolution types, the newer DLP Cinema 2K projectors will also be in use. The 2K projector has received much attention among film enthusiasts for its improved contrast and film-like deep blacks, representing a significant step forward in quality. The servers that will be used tell a different story.

Digital cinema servers are sharply dividing into two camps: the 'DCI Compliant' camp and the 'MXF Interop' camp. The MXF Interop camp (MXF stands for Material eXchange Format) was formerly named MPEG Interop. As its name heritage suggests, the MXF Interop group works to combine interchangeable MXF digital wrapping technology with MPEG2 compression. Their goal is to create a common interchange format that MPEG2 servers can deploy. The DCI Compliant camp (DCI stands for Digital Cinema Initiatives), however, does not represent an organized group. The server vendors who fit into this category are working to implement wavelet compression solutions with the desired MXF digital wrapping technology. Wavelet technology is favored by DCI, which has selected a yet-to-be decided configuration of wavelet-based JPEG 2000 as the compression of choice for digital cinema. The goal for each group is to eventually meet SMPTE (DC28) and possibly ISO (JPEG 2000) digital cinema standards, when such standards are stable enough to implement.

From a quality perspective, there is a difference between these two camps. Tests performed by Digital Cinema Initiatives dictate the use of 12-bits per color for motion picture images. Those striving to meet DCI specifications with wavelet compression are capable of 12 bit color. MPEG2 servers, however, as required by the MPEG2 standard, are limited to 8-bit color. While several people have commented to me that the difference is not visible, I cannot agree. Having just been shown a demonstration of known high quality material in both 8-bit and 10-bit color on a 2K projector, I have to say that more detail was visible with 10-bits. Numerically, 10-bit color has 4 times as much gray-scale definition as 8-bit color. DCI's tests place the limit of human vision at above 11 bits per color, justifying their rounded 12-bit color requirement. Eventually, these two camps will converge on the final SMPTE/ISO solution, capable of reproducing not just 2K images, but also 4K images.

During my visit to Industrial Light and Magic, Rick wanted me to see work in progress for Episode III. The digital clips I saw not only looked great, they looked better than Episode II. I am not the only observer of these clips to make this comment. It could have been due to the 2K projector, to the greater than 8-bit color, or to the improvement in digital work flow that ILM continues to advance. While the digital images I saw may not be possible for everyone to see today, that day is coming. And when it does, we should remember that it all started with the push of a certain film director telling the story of a young Jedi knight...