HDV in India
Byline: Steve Mullen
Shortly after arriving in Chennai, (Madras) India, I went off to my favorite bookstore to look for video magazines from the UK. I did find the usual home theater (naturally with nada on HD) and camcorder magazines. But this time I also encountered a new Indian A/V magazine that had reviews of several plasma displays now on sale in the country. After looking through both U.S. and UK Mac magazines, I found an audio magazine with a story on climbing Mount Rainier with a JVC JY-HD10. It sparked my interest in writing a story on shooting and editing HDV in India. Because I had brought a JVC GR-HD1 along with my trusty 800MHz G3 iBook, I had the necessary technology.
Before I left for India I had thought I’d get a full week in Bangalore, my favorite Indian city. Bangalore, located in the south of India, is the city where computer technology has exploded over the past decade. I expected to have time to shoot a short piece on the “New India.” Unfortunately, after two family weddings, the Bangalore trip was reduced to only a few days, ruling out any serious shooting. Thankfully, at one of the weddings I had shot a traditional wedding band that captured my interest because of its very complex drumming.
The four-person group was seated on the floor and lit by both blue fluorescent lamps and indirect daylight. The natural light came from one side, while the interior illumination came uniformly from the ceiling. The result was a nearly perfect – for the HD1’s small high-density CCD – low-contrast situation. Because the light level was not high, the camera automatically chose an exposure of 1/30 to 1/60 second. This meant that I had two exposure-control options. I could use the camcorder’s Exposure button to lock exposure while also, as necessary, biasing the exposure lighter or darker. This is the shooting technique I recommend in my HDV Shooting Guide. Naturally, I didn’t follow my own advice.
Instead, after noting the auto-exposure was spot on, I simply set the shutter-speed to 1/30 second. Doing so accomplished two goals. First, using a slow shutter thereby minimized the aperture, so depth-of-field was maximized, which made achieving focus slightly less demanding. Second, it kept eye-tracking artifacts to a minimum. (See “High-Rate Progressive,” Video Systems, Feb. 2004.)
The same lighting illuminated a nearby white plastic chair, so I manually white-balanced the HD1 on it. A quick check of the viewfinder revealed a slightly warm image – thankfully free of any blue or green tint. I quickly decided the warmth was an appropriate “look” for the subject matter. (The LCD “reads” warm so the video was not too warm.)
Because the overhead lighting was not super bright, I was able to use the camcorder’s 200,000 pixel LCD. (I’ve yet to use the camcorder’s tiny viewfinder.) For close shots, I used manual focus. I had no problem achieving focus – although it did take 5 to 10 seconds of effort. When focus is achieved, the image “pops” in the LCD, so the time required was primarily used in locating something not moving. This is important because at 1/30 second, moving images have an inherent blur. For wide shots, I simply engaged the AF system for about 10 seconds.
Because I wanted stereo audio, I used the HD1’s internal mic. While shooting, I often positioned myself very close to one of the other drums. Because they were very loud, I was worried audio might be clipped. The GR-HD1, unlike the JY-HD10, has no audio-level indicator. So I had no way of knowing until I later converted MPEG-compressed audio to AIFF files. The conversion software tallies the number clipped samples. JVC’s AGC prevented any clipped samples.
I shot about 20 minutes of HD video, just perfect for working with my iBook’s half-full 20GB disk. Although every hour of DV requires 13GB of storage space, an hour of HDV amazingly requires only 9GB. Thus, my 20 minutes could be captured and still leave more than 6GB free.
Final Cut Pro 3.0 was already installed on the iBook. And I had brought a copy of my HDVcinema Pro bundle on CD-ROM. After installing this software, I was ready to edit. However, I didn’t have time to begin editing until a few days before I returned to the United States. After the weddings and the trip to Bangalore, it was time to help my in-laws get settled in a new flat. Although this didn’t offer an opportunity to shoot video of the New India, it did provide several experiences. Many of you may have heard horror stories of its taking months to obtain phone service in India. That was in the “Old India.” One week from the call-in request, a phone was delivered at the promised hour. The desk phone uses CDMA technology to deliver Wireless Local Loop service that costs about the same as a landline.
In the United States we have our own horror stories about waiting for the cable guy. From the time of the request for Chennai cable service to the installation – 24 hours. And, the guy showed up at the requested time! About 60 channels are available, including CNN, CNBC, HBO, Discovery, and National Geographic. Many hit USA network shows are also available. Naturally, no HD programming is on hand, although when I visited the local Sony World, they were marketing a 57in. 16:9 HDTV. (In India, HD DVDs will likely be the path for high-definition programming.) On a similar note, Chennai’s Real Image and Prasad Film Group (www.real-image.com/digital.asp) are cooperating in marketing a digital cinema system. The system, the same used by the Landmark theaters in the United States, uses Windows Media 9 and plays frame rates from 24fps to 60fps. Good news for those shooting 720p30 HDV.
To capture HDV, I simply connected a FireWire cable between my HD1 and the iBook. The use of FireWire, plus the fact MiniDV tape is used, makes it easy for folks to think of HDV as simply “High-Definition” DV. There are, however, significant differences that make HDV editing very different from DV editing. Although these differences will be smoothed over by coming generations of software, they will not disappear. There are three crucial differences. First, HDV uses inter-frame compression, while DV does not. Second, HDV audio is MPEG-2, not PCM. And third, the number of pixels is either 2.7X (1280×720) or 4.5X (1440×1080) the number of pixels employed by DV.
All these differences lead to the need for a huge amount of computation to accomplish editing. This need can be satisfied in several ways. One alternative breaks the task into less computationally intensive steps, which is the approach used by HDVcinema. HDVcinema uses a technique many of us used with Premiere 4.2 when our Quadra systems weren’t powerful enough – proxy video editing.
Aspect HD from CineForm takes a very different approach – transcode MPEG-2 to another format (Wavelet) that can be processed far more efficiently. This approach enables more than four HDV streams to be edited in realtime.
MediaStudio Pro LE MPEG Edit Studio Pro LE v. 1.2, a “native” HDV editor, restricts the number of HDV streams to two and works directly with a captured MPEG-2 transport stream. With a powerful enough computer, this approach seems optimal to me. One has to wonder if Apple and/or Avid will take this approach.
Like all current approaches to HDV editing, you begin by using a standalone FireWire capture application. (I used Apple’s DVHScap.) Unlike DV capture utilities, no onscreen sound or image is available from these capture applications. With the JVC camcorder, you use the LCD (or an NTSC or HD monitor connected by a component analog cable).
A serious shortcoming of the JVC camcorders is that no timecode is sent through FireWire with the MPEG transport stream. That means “marking” a single segment for capture is not possible. Nor is batch capture. Even worse, because it makes using Apple’s OfflineRT codec more difficult, batch recapture from an edited production is not possible.
Once I captured the 20 minutes of source material, I used the HDVbridge utility in HDVcinema to “demux” the transport stream file into MPEG-1 Layer II (MP2) audio and MPEG-2 video files. Unfortunately, while FCP will import MP3 files, it will not import the MP2 files used by HDV. That means an HDVcinema-supported utility must be used to convert MP2 files to AIFF files. Clearly, with adequate computer power, both demuxing and audio conversion could be performed during the capture. And this is exactly what XtractorHDV supplied in the Indie HD Toolkit from Heuris does.
After demuxing by HDVbridge or XtractorHDV, the resulting MPEG-2 video poses a problem. In a white paper, Heuris states, “Final Cut Pro does not natively read MPEG-2 content. This means that in order to perform realtime editing, the sequence will have to be rendered to a format that is native to Final Cut Pro. For best quality, we recommend using no compression.” Unfortunately, each 60-minute cassette will require 300GB of hard disk storage. Using this technique, my 20 minutes would have required 100GB.
Thankfully, Apple already provides a solution – OfflineRT compression. I used an HDVcinema Pro supported utility to batch process the MPEG-2 files to OfflineRT files. In the future, FCP could support this task during FireWire capture – as long as the Mac had sufficient CPU power and disk bandwidth. This is an ideal time to do so because the decoded audio can be placed directly into the OfflineRT file. And of course, there’s no reason the capture/demux function can’t be integrated into the NLE itself.
After importing audio and video files, I edited a one-minute piece exactly as I would using DV. Then I replaced the OfflineRT source files with the original MPEG-2 files. Next, I exported the timeline as an uncompressed movie. (The movie used 5GB.) Heuris has another approach. Its XportHD QuickTime plug-in encodes an MPEG-2 transport stream directly from an FCP timeline, thus dramatically reducing movie storage requirements. And while its XtoHD utility will record this file to D-VHS, it will not record back to an HDV camcorder or the new JVC portable HDV deck.
Now, using additional HDVcinema-supported tools, I encoded the uncom-pressed movie to an MPEG-2 video and then converted it to an HDV transport stream. Next, I used DVHScap to transfer the transport stream, via FireWire, back to the GR-HD1. Clearly, these functions could be integrated into an HD capable version of Apple’s Composer.
I also converted the MPEG-2 Time-line to a DV Sequence and recorded it via FireWire to the GR-HD1. When I returned to home, I also recorded the HD video to D-VHS. If I’d had a PowerBook with a SuperDrive, I could have used HDVcinema tools to burn a widescreen or letterboxed DVD without using DVD Studio 2 or Toast.
To show my one-minute masterpiece, I located a multi-system TV – an easy task in India – and connected it to the HD1. By selecting the “ALL TO 480i” to a “4:3 monitor” modes on the camcorder, my movie played back perfectly in letterboxed widescreen with stereo sound. Success!
Two things are clear to me. The first few years of the HDV revolution will find a confusing set of editing techniques marketed. And you can bet each will be promoted as the “best.” In short, HDV’s introduction will mimic that of the introduction of DV. This means a very fluid situation for buyers. For example, I’ve now released HDVcinema Partner, which is designed to work with the Heuris Indie HD Toolkit to provide a comprehensive solution. These packages will likely – at some point in the future – be replaced by solutions from Apple and Avid. Avid, for example, has already announced it will support the critically important Windows Media 9 HD encoder.
But more important, it’s clear that even at this early stage of HDV postproduc-tion – whether you work with a 3.4GHz Pentium 4 PC or an iBook – HDV editing can be accomplished. And from the folks I talk with, accomplished in a highly profitable way.
To comment on this article, email the Video Systems editorial staff at email@example.com.
COPYRIGHT 2004 PRIMEDIA Business Magazines & Media Inc. All rights reserved.
COPYRIGHT 2004 Gale Group