Following up a bit on my last post, the photography website, Luminous Landscape, has a brief user review of the new JVC MC500 camcorder that records direct to compact-flash card, rather than DV tape. Now, before I get branded a luddite, I believe that this is the direction which digital video is going and should go. However, I also have to say that it’s just a bit early yet.
This particular camcorder is more of a proof-of-concept for early adopters rather than something useful for the avid amateur or pro videographer. The main problem is that it records in MPEG2, the same format used on DVDs. Now, MPEG2 is a great delivery format, it allows you to cram a lot of high quality video into just a few gigabytes. But to do so, you sacrifice a lot of flexibility.
One thing you sacrifice, which is very important to videographers, is good frame access. That is, unlike conventional film, and even higher-level digital video formats like DV, MPEG2 doesn’t record full frames. Instead it records a full frame, a “key frame,” every so often — typically anywhere from every 5 frames to every second. Now, there are 30 frames of video per second in the US NTSC standard.
After the key frame, every additional frame of video isn’t full. A bunch of data is left out. The goal is to record only data that has changed. So, in theory, if the video is of a ball that doesn’t move, the key frame is picture of the ball, and each subsequent frame has no information, since nothing changed. If the video was of a ball rolling against a solid black backgroud, the key frame would show the whole initial scene and the subsequent frames would only have the parts that changed — the ball and what was formerly behind the ball.
Making this work in the real world is somewhat imperfect, but can yield good results. But, when it comes to editing, you might want to cut at a frame that isn’t a keyframe, so your editing computer has to go back and fill in all the information that isn’t recorded. It can work, but a lot of data can be lost.
DV format, used in miniDV camcorders, records 30 full frames a second. So you can choose a cut anywhere and get a full frame. This is a very important point. But it also means that you’re dealing with a lot more data.
The JVC camcorder gets an hour of MPEG2 video on a 4 GB CF card. That same card would only hold about 20 minutes of DV video. That miniDV videocassette you buy for $4.00 actually holds 13 GB of data. They don’t even make 13 GB compact-flash cards yet, and the 4 GB cards still cost more than $100 each.
So, you can see why tape is still very efficient with regard to gigabytes per dollar. But the downside is that tape can get jammed, and it’d difficult to make it run faster than real-time. Typically, it takes an hour to upload an hour of video from your camcorder to your computer.
In my day job of video production, we’ve begun using portable 60 GB hard drives that are designed to interface with DV camcorders to store video as we shoot. It often saves us several hours of uploading times, since we can just hook up the drive to our editing PCs by firewire and transfer the files.
Yet, we still run tapes even when we’re running the hard drive. Why? Well, have you ever dropped a hard drive six feed onto a tile floor? Your chances of data survival are pretty slim. But I’ve dropped a DV videocassette many times, and haven’t yet totally lost the data, even if I had to perform a bit of surgery.
Also, one 60 GB DV video hard drive costs about $750. So we can’t afford to record 4 hours of video and just put it on the shelf. So how can we archive all that video we shot once we’ve edited the final project? Ah ha, just keep that $4 videotape we used during the shoot.
For the moment, tape still rules, even though hard drives are helping to speed up and smooth our our workflow. But until there is a good reliable removeable media that stores at least 13 GB cheaply, tape is going to rule the roost.
And, I will admit, I do look forward to that day when tape takes a back seat.