Dispelling Some Myths About Sony's XDCAM HD

As one of several DPs working on the TV show Mythbusters since 2004, Peter Heap has shot about half of the series’ roughly 100 episodes. This includes capturing footage inside an underwater car, from a helicopter high above the Rockies, beside 100 pounds of explosives (and a burning fuse), and extremely close to live alligators, a 747, a cruise ship and frighteningly loud machine guns.
He’s one of two teams that shoots mostly in San Francisco for up to 12 months at a time before returning to his home in Sydney, Australia, where the show (produced by Beyond Productions) is written and edited on Avid systems. Once finished, an HD tape is sent to Discovery Channel headquarters in Washington, DC.

During his career Heap has used virtually every type of digital video camera. His latest is a Sony PDW-350 XDCAM HD camcorder.

Q: Explain your HD workflow and how you make use of the PDW-350’s low-res proxy files.
A: Every Friday we FedEx a shipment of discs and HDV tapes back to Sydney. So some of that material (shot at 35Mbps) will have been sitting around since Monday, some from Tuesday, etc. And sometimes we won't get back from location until after the shipment has left on a Friday and those discs will be a week old before they're sent and up to nine days old before they arrive in Sydney.

So sometimes I'll copy the proxy files off a disc (from the camera to a laptop) using Firewire and then send them via the Internet using FTP. That can buy up to a whole week for the post-production team; the off-line editing all takes place using proxies. The original discs are still sent back to Sydney so that the high-res files can be used in the on-line edit.

Q: Some would say that the 1/2-inch CCDs used on the PDW-350 camcorders are not the best for good-quality HD acquisition. What's your feeling on this?
A: The bigger the CCD the shallower the depth of field. Which is a good thing because it helps to convey a feeling of depth on a 2D screen. But while we lose the separation from the background that we’d have with 2/3-inch CCDs, we also benefit from the increased depth of field, in that focus is less critical. We shoot very much on the fly, often wide open, so focus issues are more forgiving. You've got to look for the positives sometimes. In an ideal world we'd be using CCDs that matched a 35mm frame size and would give us great separation between planes within the picture. Maybe one day.

Q: You don’t use optical discs for some of the more disruptive scenes where vibration or moisture would be a problem. Do you find the camera and discs to be robust in tough situations?
A: Generally we'll have a bunch of cameras rolling when we conduct an experiment. The XDCAM HD concentrates on the hosts, and the other cameras shoot the experiment. Our second camera operator will set up 3 or 4 HDV cameras and a Memrecam K4 high-speed camera [made by a company called NAC Image Technology, in Simi Valley, CA.]

Often these cameras will be close to something violent or explosive or loud, and it’s not worth risking $40,000 cameras when you can get adequate shots from $1,000 cameras. Usually those shots will only be on screen for a few seconds at a time, so we feel we can get away with using consumer cameras. We’ve incinerated, shot and smashed a few cameras on the show and that’s in spite of trying to protect them. Sometimes things happen that we just don't expect.

Q: Has the migration from SD to HD been a smooth one?
A: With the change from SD to HD we stopped using lipstick cameras and started using HDV camcorders. One of the reasons was cost. We couldn't find affordable HD lipsticks and recorders, and it’s also much quicker to set up a camcorder (with one battery) than it is to set up a lipstick (with one battery) feeding into a recorder (with another battery) and all the cables between the two.

However, the change to HD hasn’t been without problems. If an SD camera took a knock you’d get breakup on one or two frames, and then it would keep on recording. But the HDV cameras actually stop recording for a moment and then start again. This leaves us with video of the experiment about to happen and video of the aftermath but the actual impact isn’t there. Not very good for a show like ours!

It seems that what's happening has got something to do with the inter-frame compression. Each frame shares information with what's called a group of pictures (GOP), which is made up of 15 frames and, as far as I can tell, if you get a failure of one frame you also lose the other 14 frames. What this means for us was that if the impact occurred on the 10th frame in a GOP, we would lose nine frames before the impact and five frames after it.

To get around this we’ve ended up going back to an arrangement not unlike we had with the lipsticks. We record on a separate recorder if we suspect that there will be a big physical shock. When we first experimented with this setup we discovered that the HDV decks were just as susceptible to shock as the camcorders, so we tried packing them in foam to cushion them, but without much improvement.

After a bit of experimenting (we are “mythbusters” after all) we discovered that putting the record decks into Pelican cases and filling them up with packing peanuts worked best. But even that won't work if the Firewire cable comes out. When we blew over a taxi with a 747, we had two HDV cameras mounted inside the taxi for POV shots. The recorders were packed in their cases with peanuts and the cases were then placed in the trunk and surrounded by big bags containing yet more peanuts. Everything was going great until the taxi was hit by the blast from the 747’s engines and the trunk lid was ripped off its hinges. The cases were picked up by the 55,000 lbs. of thrust and were hurled a hundred feet down the runway. One of the myths we busted that day was that a FireWire cable is stronger than a jumbo jet.