Archives: November 2007

25/11 Beowulf 3D: some thoughts

Robert Zemeckis' new CGI extravaganza is out, in virtually every cinema format known to man: film, digital 3D and IMAX 3D. The digital 3D format is (like virtually everything 3D at the moment apart from the IMAX) Lenny Lipton's RealD system, which uses circularly-polarised glasses and an liquid crystal polarisation-flipping shutter (the "Z-screen") to display alternate frames from a digital projector running at 144fps, to each eye alternately. (This implies that each frame of the original film, which was rendered at the industry-standard 24fps, is displayed three times to the eye it was intended for, at 72fps; this is a projection technique known as "triple-bladed shuttering".)

A group of nine of us went to see the film at the Odeon in Wimbledon; here are some of my observations.

  • Clarity: Digital projection is the future. It was so nice not to have to watch a film through a haze of flickering dust and reel-change blobs!

  • Headaches: After 2 hours of stereoscopic viewing, only two out of nine of us had headaches. That's actually pretty amazing in 3D cinema terms, I believe.

  • Stereo flicker: The ads and trailers before the show were projected from a film print. Projection switched over to digital when the stern message from our country's noble guardians of copyright appeared, followed by the BBFC certificate for the film. On these static images, mostly white on a black background, clear flickering was apparent to me, particularly when I moved my eye around the screen. This flicker was noticeable to me in high-contrast scenes throughout the first half of the film, but by the time it got to the credits, I was barely able to perceive it. Because the image always flickered when I moved my eyes around the image (a process known as "saccading"), the implication is that it was being caused by some interaction between the system of projection and the behaviour of the Human Visual System (HVS) during the saccade. It is well known (eg see that Wikipedia article) that "saccadic masking" occurs during these movements: the HVS "turns off" its response to low spatial frequencies of monochromatic light. When viewing a 3D film projected with the RealD system, after each saccade, one eye has to wait longer than the other before it sees any light on the screen. While the interval is short (~1/144s), the human eye and the HVS are not used to operating in environments where this sort of thing happens, so it is conceivable that the HVS interprets this as flicker until the brain trains it not to notice it. (And never forget that your ability to interpret the flat, flickering, perspective-projected images of 2D film and television is itself entirely unnatural, but since you were probably brought up with at least one CRT in your house from birth, you don't remember the learning process.) There is probably a threshold of perception too, so increasing the frame projection rate might well render the effect less visible. It would be very interesting to hear of any research that's been done on this issue.

  • Judder: I found the judder in this film particularly objectionable. It was particularly obvious in scenes where faces were moving - it was common for the face to be unviewable during the movement. A colleague who saw the film with me thinks that artificial motion blur has been added to reduce the prominence of this artefact, but often only for the middle part of the motion, ignoring the ends. It would be interesting to step through the frames on an HD-DVD or Blu-Ray version of the film when available, and see what really went on. It's also not clear whether or not viewing a film in stereo increases the perception of judder compared to a 2D projection - I have a hunch that it does, but no scientific evidence, so the judder in Beowulf could just be a consequence of poor motion blurring during the rendering process and the use of triple-bladed shuttering in the projection of the film.

  • Throwing things at the camera, and making sharp pointy things stick out of the screen and into the audience: Please directors, just stop this already. You're supposed to be making a film, not a theme park ride.

Overall, my gut feeling was that most of the technology problems have been solved. If directors can keep the 3D gimmickry down to a level where it doesn't interfere with audiences' immersion in the film (and that's a big "if" - Lenny Lipton says lots of diplomatic things on the subject here, for what it's worth, but this bit of trivia on IMDB hardly fills me with confidence), I think that 3D cinema could become a long-term mainstream proposition.
  • Time: 10:30PM
  • Category: BBC