Choosing a Camera for Online Video Production
How much does this matter? In my resolution chart tests, formats such as HDV and DVCPRO HD proved incredibly adept at recreating the information that was thrown away during the storage process—the difference in resolution is almost imperceptible. Still, all other things being equal, you’d prefer a format that stores the video data at the same resolution that it’s displayed.
For example, most early AVCHD camcorders stored the captured video at 1440x1080 resolution rather than the maximum of 1920x1080. Though the difference is modest, if buying for professional use, you should opt for a full-resolution AVCHD camcorder (look for the "FullHD" logo) over one that stores at 1440x1080.
It’s also instructive to understand the compression technology used by each format and the associated data rates, which affect quality, editability, and storage requirements. Historically, DV uses an I-frame-only encoding scheme, which retains very high quality and is relatively easy to edit. Interestingly, DVPRO HD uses the same basic scheme, but the data rate is four times higher, which neatly corresponds with the 4X increase in stored pixel resolution (345,600 compared to 1,382,400). For this reason, DVD PRO HD also offers very high-quality video and is easy to compress, though the 4X DV data requires capacious storage.
HDV uses long-GOP MPEG-2 to encode HD video down to the same 25Mbps data rate as DV. long-GOP means a mix of I, B, and P-frames, or a compressed structure that’s harder to decode and edit than I-frame-only DV; this is because the majority of frames need to refer to other frames in order to construct an entire frame of image data. In the Core 2 Duo and beyond era, this really isn’t a problem, and editing is very efficient. Ditto for XDCAM HD, which uses the same basic technology at up to twice the data rate.
However, AVCHD uses long-GOP H.264 for compression, with a maximum data rate of 24Mbps, which results in a higher-quality file than the long-in-the-tooth MPEG-2 but may be much harder to edit depending upon your program and platform. For example, Final Cut Pro converts AVHCD to ProRes during ingest, which costs you some conversion time but makes editing much more responsive. Adobe Premiere Pro CS4, by contrast, edits AVCHD natively, which is very memory-intensive. On a 32-bit Windows system, which can only address 4GB of memory, you may experience some editing delays with even moderately complicated projects, and rendering is very slow. However, on the Mac and on 64-bit Windows systems, which can address much more memory, performance is reasonable.
Now that you know all this, when do you care? At a high level, all the formats have been tested and have proven worthy. They’re all very good. In extensive comparison tests, which unfortunately didn’t include XD CAM HD, I saw very little quality difference unless the clips included extraordinary (above and beyond the call) levels of motion. If I had to rank them, I would choose DVCPRO HD first, AVCHD second, and HDV third. Still, the difference is so exceedingly minor that I wouldn’t use this factor to choose between the camcorders.
However, when comparing AVHCD camcorders, be sure to identify both the maximum storage resolution and the maximum storage data rate. The Panasonic HMC150, which is the only AVCHD camcorder I feel comfortable recommending for professional use, was the first to record at 1920x1080 at a 21Mbps data rate. All early versions of AVCHD camcorders recorded at 1440x1080 resolution or lower, with a maximum data rate of 14Mbps.
The Panasonic HMC150, which is the only AVCHD camcorder I feel comfortable recommending for professional use, was the first to record at 1920x1080 at a 21Mbps data rate.
In addition, when comparing camcorders within a class, such as HDV camcorders, knowing the storage resolution will help you evaluate the quality of the CCDs or CMOS sensors used by the camcorder, as we’ll see in the next section. Finally, formats are inextricably tied to storage medium, which can be a crucial decision factor.
Question Three: How Many Pixels in That CCD Thing?
The next question to ask is how many distinct pixels the camcorder’s sensing devices actually capture. Why is this important? Many early HDV camcorders had sensors with fewer than the 1,555,200 pixels necessary to capture the stored 1440x1080 image, pixel for pixel. For example, the HVR-Z1U camcorders that the Z5U will replace only had about 1 million effective pixels. This means that the Z1U captured 1 million pixels and interpolated upward to the required 1,555,200. Quality was good, but it was never as sharp as camcorders such as the Canon XH A1, which were built with 1.6 million-pixel CCDs and could capture the image without interpolation.
Canon was the first to step up with full-resolution CCDs; Sony appears to be following with the HVR-Z5U. Specifically, the product’s press release states, "The CMOS sensors capture full HD 1920x1080 resolution, resulting in better picture quality when recorded onto miniDV tape in the HDV format (1440 pixels x 1080 lines)." I haven’t compared HDV systems that captured at 1920x1080 (such as the new Sony) with those that captured at 1440x1080 (such as the Canon), so I don’t have an opinion regarding which approach is superior. I do know that with all other factors being equal, higher pixel counts in the image sensors improves sharpness.
The new Sony HVR-Z5U is reportedly the first Sony HDV camera with sensors equipped to capture full 1920x1080 HD resolution.
Panasonic is the only vendor not drinking the pixel-count Kool-Aid. The DVCPRO HD-based HPX-170 stores at 1280x1080 resolution, but its three 1/3" CCDs have 960x540 active pixels and use a technology called "pixel-shifting" to produce the necessary additional pixels. In my tests, the results were slightly softer than the XH A1, which means less detail. Similarly, while Panasonic hasn’t released the CCD pixel resolution for the HMC150 camcorder, it has stated that the effective pixels are fewer than the 2,073,600 necessary for full-resolution capture, again, using pixel-shifting to fill in the difference.
With both camcorders, the difference was only noticeable in resolution chart testing, with real-world images very sharp and clear. Still, if I were comparing AVCHD camcorders, I would expect models with higher pixel counts to produce sharper video than those with lower pixel counts—not dispositive, but it’s definitely a good data point to identify.
Question 4: What’s the Storage Mechanism?
Next, consider how the camcorder stores the video: both the primary media as well as any ancillary options. For example, most HDV camcorders store video on tape and include FireWire output for simultaneous or alternative capture to a hard disk recorder, such as the HVR-DR60 60GB hard drive that Sony offers as an option to the HVR-Z5U. However, most AVCHD camcorders only have a single SD card slot with no FireWire option for ancillary storage—a bummer if you’ve already purchased an HDD option and want to supplement your supply of SD cards. This section discusses some considerations for the most common media options.
Most HDV camcorders store include FireWire output for simultaneous or alternative capture to a hard disk recorder, such as the HVR-DR60 60GB hard drive that Sony offers as an option to the HVR-Z5U.
Tape is tried and true for HDV, and it represents a cheap archive system for the source footage. On the other hand, capture from tape occurs in real time, which is slow, and tape mechanisms are fragile devices and usually the first subsystem to fail on a camcorder. Recording to tape means occasional dropouts, or short sections of lost data, and finding scenes on tape for playback is cumbersome.