THE PHANTOM MENACE, DIGITAL PROJECTION AND YOUR HOME THEATER by Dan Ramer

No...there...is...another...

Why did George Lucas decide to put electronic projection of Phantom Menace in two theaters in Northern New Jersey and two theaters in Los Angeles? Wouldn't it have made more sense to put one projector in each of four widely separated locations to give more people the opportunity to see a filmless theatrical presentation? In response to my piece on the Hughes-JVC, I received e-mail from people who were reporting artifacts that couldn't possibly have been generated by Hughes-JVC projector I saw in Paramus. My correspondents had to have seen a different projector.

Then it occurred to me. Lucasfilm must have installed a different projector in each of two theaters on each coast. (Perhaps Lucasfilm was interested in the public's reactions to two different projectors in the same demographic region.) The Texas Instruments Digital Light Processing (DLP) projector (the other contender for theatrical projection) must be in the Secaucus theater. I called, and sure enough; they're using the TI DLP. I sent an e-mail to Doug Darrow, the TI representative for this project, with a link to my Hughes-JVC article. (If you missed it, you may find it here.)

I asked if I could solicit the information I might need to prepare a piece on his projector. He explained that he had seen my piece on DVDFILE.COM before I e-mailed - someone had sent the link to him - and kindly volunteered to offer any appropriate assistance. So once again, I'm off to The Phantom Menace, this time in the Sony Meadow Six Theatres located in Secaucus. When I arrived for Sunday's first showing, I was amused that either the theater or TI went out of its way to make this experiment a festive occasion. There was a little group of Dixieland Jazz musicians entertaining the folks on line. Once inside, as they entered theater one, moviegoers were given digital theater "backstage passes" to wear around their necks. Nice touches.

I carefully selected my seat so that my field-of-view was about the same as for both the Hughes-JVC and the filmed presentations I had seen before, approximately 54 degrees. When the show began, it was immediately clear that this would be a very different experience. TI's color balance was closer to film that the Hughes-JVC; it was a little more subdued. TI's contrast did not appear to be quite as high as the Hughes-JVC and objects didn't have the same illusion of depth (but the numbers would seem to contradict this observation). The convergence was perfect; I never observed the slightest color-fringing characteristic of RGB misalignment. I couldn't find any interlacing artifacts, either in the motion of projected objects or in the telltale presence of nearly invisible horizontal intensity banding running down the image at about one picture height every four seconds. Not one dead pixel was apparent. I saw no blooming.

During our first glimpse of the Naboo Palace courtyard as the Battle Droids enter to occupy the city, the ground was almost uniformly white and featureless as projected by the Hughes-JVC. On the TI projector, the ground showed texture. (I did spot one odd instance of blooming, although I suspect a source problem. It was during the scene in which R2-D2 meets C3-PO for the first time. On his domed head, R2-D2 has a sensor just below his eye. It frequently changes from blue/white to red. When it did, the red bloomed (in that scene). But the intensity of this light was so much lower than many other objects that ran through the film, I had to conclude that it wasn't the projector.)

But once again, I cannot report a presentation without flaws, and this one was rather serious. The TI projector is based on DLP technology, a dense array of Digital Micromirror Devices. The DMD array is in the aspect ratio of 5:4 and measures 1280x1024 pixels. To accommodate this array, the Panasonic HD M-JPEG telecine process was specifically tailored to the TI presentation. The total width of the film was adjusted electronically to occupy only 1280 of the 1920 pixels available. The total height was adjusted electronically to occupy only 1024 of the 1080 pixels available.

When projected, the aspect ratio was restored onscreen using an anamorphic lens to stretch the frame horizontally. So 1280 columns of pixels were projected across a screen that was about fifty feet wide, resulting in columns about one-half inch wide. (The pixel rows were about a quarter inch in height and were not visible.) From where I was sitting, the vertical columns of pixels were quite apparent. And having one-third fewer pixels across the screen tended to soften the image horizontally, reducing the level of detail. In the opening shot of the pod race arena, the very long shot with tiny figures walking about, the figures' moving legs were clearly visible as projected by the Hughes-JVC, but were indistinct as projected by the TI.

After the show, I met with Scott Dewald, a member of Texas Instrument's on-site staff. He invited me into the projection booth for some Q&A. As with the Hughes-JVC presentation, the film was digitized under the supervision of Lucasfilm's THX Division and stored on a Pluto RAID array (disk drives with redundancy). From the RAID array, the data was passed through a Panasonic HD M-JPEG decoder, then to the TI DLP projector. I mentioned that I hadn't spotted any motion artifacts and asked how the 3:2 pulldown flaws were removed. It was explained that circuitry within the TI projector reverses the 3:2 pulldown, reduces the data to the original film's frame-rate of twenty-four frames per second, and projects it at that rate.

As in Paramus, synchronized, convention film was run in parallel to permit almost instantaneous switchover should an electronic element fail. And I noticed a spare TI projector in the booth, a conservative precaution. The TI projector's light path is somewhat similar to the Hughes-JVC. Each starts with a Xenon Arc Lamp light source, each splits the light into RGB with dicroic mirrors, each controls the two dimensional intensity of the projected image with a light-reflective process (of course, the TI uses its proprietary array of DMDs on a DLP chip). While the Hughes-JVC projects three colors and combines the light on the screen, the TI uses a prism to combine the colors and uses only one lens system to project the image. I was told that the dynamic range of the TI DLP projector is 1000:1, equivalent to the Hughes-JVC.

Yet, TI's images didn't seem to have the same illusion of object depth, that almost three-dimensional effect, I observed in Paramus. Perhaps it was the lower horizontal resolution that marred the TI's ability to create that illusion, or perhaps it was the contrast within the source material. But this resolution issue is purely temporary. You'll be pleased to read that TI is diligently working on a true HDTV DLP with a full 1920x1080 DMD array. As of today, if I had to choose between the two presentations, I would have to recommend the Hughes-JVC. When TI introduces its full resolution HDTV DLP projector, the projected level of detail should be equivalent to the Hughes-JVC. We'll then have a real race on our hands.

Since my first exposure to the Hughes-JVC, Ken Bylsma sent me a message saying that they're now looking into the blooming/convergence problem I reported. He asked if I'd like to be kept informed. I replied that I'd be very grateful and would also appreciate the opportunity to see the results in the Paramus theater should they resolve the problem(s). If they invite me back, I'll be sure to report their progress to you.

Also see the part one of this piece with Dan's first impressions of Digital Projection tehcnology. Still want more information? Read more about the JVC-Hughes impressive projector (and see a photo), an explanation of the Image Light Amplifier, and read an article discussing various projection methods.