Phantom Canyon (2006) Footage courtesy of Stacey Steers IMDb entry Flash media player by Jeroen Wijering and used under license Get the Flash Player to see this player.
|
This is a clip from Stacey Steers' short film Phantom Canyon, recently acquired for exhibition at the Hamburger Kunsthalle, Deutschland. Stacey's animation technique utilizes Xerox cut-outs and a Mitchell 35mm motion picture camera. Stylistically Phantom Canyon is very different from her two previous films, Watunna and Totem, yet they all explore the sub-conscious, spiritual connection between humans and animals. I asked Stacey about the symbolism of a particular recurring motif in her latest (and darkest) film and she simply responded that she works on a purely intuitive level; she doesn't really analyze it in such terms. Stacey Steers uses traditional motion picture equipment and processes to create her films, but - like every other filmmaker - she needs also to have various video and digital versions of her work. Accordingly, her timed 35mm InterPositive (IP) print was transferred 'one-light' (D-min to D-max) on a Cintel C-Reality CRT-based film scanner to HDCam 1080p24 format. The show was pillarboxed within the HDCam's 16x9 frame, retaining the intended 4:3 aspect ratio. The HD telecine material arrived at GW Hannaway & Associates (GWH&a) for final HD 24P mastering, subsequent down-conversion and progressive MPEG-2 encoding (for DVDs). Then in November 2006, the Hamburger Kunsthalle requested of Stacey a PAL (phase-alternating line) format Digital BetaCam version for inclusion in the German museum's collection. This sparked a discussion among Mrs. Steers, GWH&a and the Hamburger Kunsthalle. What was the best method of converting and delivering a PAL version? There are many different technologies and philosophies for video standards conversion. With regard to film-originated North American video to PAL video conversion - and without belaboring all the arcane engineering details and permutations - there are basically two methods:
It should be noted that the Teranex Xantus parallel video computer running its imageConvert™ software has some very sophisticated techniques for converting to/from PAL in real-time. Some of these features include: broken film cadence correction, 3:2 pulldown detention & removal, and a 'slow PAL' video signal output for maintaining 2:2 film cadence over the playback's 50i signal. (Avoiding the creation of 'hybrid' frames.) Further, it has standards-compliant color space and gamma conversion included. I considered using the Teranex, but opted instead for Apple Shake since the material was B&W. For Stacey's film, I felt the best way to perverse the intricate nature of her beautiful animation, while also delivering a technically correct video to the museum, was to maintain the film cadence. The only trade-off with this method is that film must be sped up 4% from 24 fps to 25 fps. [Actually, to be precise, there is a 4.27083% speed-up from 23.976 fps to 25 fps.] To the human eye, this is typically not noticeable. So the visual conversion (all progressive images, none interlaced) was simply: | speed up by 4.27083% | \/ 1920x1080 at 25 fps | crop 16x9 to 12x9 (4:3) | \/ 1440x1080 at 25 fps | resize to PAL resolution | \/ 720x576 at 25 fps As previously mentioned, such a speed change is not noticeable visually. However, the affect on sound is severe; altering its pitch by nearly a semitone. (0.724 semitones higher.) So once the picture is converted to PAL, the more interesting issue is: changing the film's audio to the new (shorter) TRT while maintaining the original pitch. In the broadest terms, this is an audio resampling problem. This is where Apple's Soundtrack Pro's 'time-stretch' feature came in handy. Today, these are very common DSP tools, available to anyone with a computer. (The RTAS plug-in Serato Pitch 'n Time is an excellent tool for Pro Tools users.) The trick, however, is in knowing exactly which parameters should be fed to the digital signal processors for a given situation. Most of the audio pitch-shifting and time-stretching programs allow the user to alter the audio by inputting one of the following sets of parameters:
Both the AES (Audio Engineering Society) and the SMPTE (Society of Motion Picture Engineers) have standardized the NTSC video-audio relationship in terms of sample count as follows: 8008 audio samples per every five NTSC video frames. (Fs = 48 kHz) For all video/audio permutations, see Table 1 below. In order to correctly time-stretch Phantom Canyon's soundtrack without pitch change, we must calculate the exact digital audio sample counts of both the original '24P' and the new PAL versions. See Table 2 below for these numbers. The prescient Hamburger Kunsthalle agreed to accept an uncompressed PAL format Quicktime file as the deliverable, instead of the originally requested PAL Digital BetaCam. Everyone immediately saw that Quicktime file would be a higher-quality master. Further, the 12 GByte Quicktime was losslessly data-compressed via GZIP to about 3 GBytes - enough to fit on a DVD-R data disc. These steps saved Mrs. Steers hundreds of dollars in delivery costs. That's how it's done. 2009 Update: Apple Compressor and Telestream Episode Pro now can directly handle the audio correction for speed-up or speed-down transcoding. It has been a great honor having Mrs. Steers entrust me with her cinematic sights and sounds. I am also very fortunate to have gained her friendship. Visit the MindCine main site. |
|
|