Affordable HD Video Format
- Categorized in: Choosing a camcorder
If you’re considering buying a new affordable HD camcorder, you should have at least three formats in mind: tried-and-true HDV, up-and-coming AVCHD, and the ever-more-affordable DVCPRO HD. By a stroke of good fortune, I have three such camcorders in hand right now, and I thought it would be a good time to discuss their relative merits. Specifically, I’ve got the Canon XH A1 ($3,299 street), the Panasonic AG-HPX170 ($5,195 street), and the brand-spanking-new Panasonic AG-HMC150 ($3,495 street). Granted, the HPX170 is a much more flexible camera targeted at higher-end markets than the other two, but it will at least be on the radar screen for folks comparing the other two.
While I'm on the subject of caveats, I’ll make the obvious observation that a camcorder is different from the storage format and that you can’t make conclusive observations about a format by looking at one camera. That said, at least for AVCHD and HDV, these are two highly credible cameras, if not best of (respective) breed.
I’m particularly excited about the AG-HMC150, which is the first AVCHD camcorder I’ve reviewed that can capture at full 1920x1080 resolution at up to 24Mbps—about the same as HDV and the highest “legal” AVCHD data rate. Although I’ve been an advocate of both the format and the flash-storage mechanism, it feels as though AVCHD has been hamstrung both by lower-resolution capture and 14Mbps storage rates. If AVCHD is going to really fly in the prosumer space, it will be with a camcorder such as the HMC150.
To be clear, however, my focus here is not so much on the cameras—that will be the subject of two separate reviews for the two new Panasonics—but the formats. In the first installment, I’ll discuss the formats and detail their editability. Next installment, I’ll look at quality. I’m not going to compare or contrast the P2 vs. tape vs. SD media comparison at all in either piece, but I will touch upon them in the individual camcorder reviews.
Distinguishing the Formats
Let’s start with a deeper look at the formats. Table 1 identifies the formats and features that I’ll focus on, starting with the formats' aspect ratio. I’m going to run through this quickly because I detailed many of these considerations in an article for Digital Content Producer.
By the term "aspect ratio," I mean the relationship between the number of pixels stored in the digital file compared to the pixels displayed upon decompression. For example, as you probably know, the HDV format actually stores 1440x1080 pixels in the file, but it displays 1920x1080, zooming the stored pixels by 1.33X to achieve the wider display.
AVCHD has a variable aspect ratio. Although, as mentioned, the AG-HMC150 stores 1920x1080 pixels in the file, so the storage to display aspect ratio is 1:1. This means that each displayed pixel is represented by a full pixel in the stored file, which is the best possible scenario for quality. In contrast, DVCPRO HD stores 1280x1080 pixels in the file, zooming the horizontal pixels by 1.5X during display to 1920x1080.
Of course, quality is also affected by the compression format—or codec—used to store the file, as well as the associated data rate. Again, I detail the differences between the codecs in the above referenced article. However, to set the table for this discussion, I’ll note that HDV uses long-GOP MPEG-2 video at a data rate of 25Mbps, while AVCHD uses long-GOP H.264 at a data rate of up to 24Mpbs.
Most folks who have actually compared MPEG-2 with H.264 will tell you that H.264 delivers better quality but is a more complex format that’s harder to decode—which, at least in theory, translates to less editing responsiveness when you're editing natively. In contrast, the I-frame-only structure employed by DVCPRO HD is clearly less efficient from a compression standpoint—which is why it requires 100Mbps, but it should also be the easiest to decode.
Table 2: Comparative processing on an HP xw6400 workstation.
Because most of us have to actually edit the video that we shoot before we can charge our clients, I wanted to next focus on the editability of each format. I’ll note up front that this varies by editor. For example, Apple Final Cut Pro converts AVCHD to ProRes 422 during import, while editing HDV and DVCPRO HD natively. That neatly negates AVCHD’s inherent complexity. It was hard to get a true apples-to-apples comparison, but what I did find didn’t convince me that AVCHD would be extraordinarily more difficult to work with than either HDV or DVCPRO HD—at least if you had a very competent computer.
Before reaching this conclusion, I ran several tests on two eight-core HP workstations. On the HP xw6400—running Sony Vegas with the DVFilm Raylight MXF plug-in—I loaded two 15-second DVCPRO HD streams on the timeline, the top stream configured at 50-percent size. I then previewed the project and recorded the CPU load during preview on Window Task Manager.
Next, I added several effects to the top track, spinning the picture-in-picture 360 degrees from start to finish, while expanding from 50-percent to 100-percent resolution, and threw in a moving lens flare for good measure. Then I duplicated the tests for AVCHD and HDV. The results are shown in Table 2.
As you can see, DVCPRO HD actually required more horsepower to preview and took the longest to render, which was a surprise. Although AVCHD was harder to preview, it rendered about 10 percent faster than HDV, another surprise. None of the numbers were scary though, and the overall grouping was relatively close.
Next, I ran a similar set of tests on an eight-core HP xw8400 workstation, focusing on the Adobe family of products. Premiere Pro CS3 doesn’t support AVCHD, so I ran AVCHD tests on a beta (but announced) version of Premiere Elements on the assumption that if the next version of Premiere Pro does support AVCHD, it will use the same codebase. I tested the other two formats in Premiere Pro CS3, because Premiere Elements doesn’t support MXF.
Although AVCHD proved least efficient in these tests, again the difference wasn’t exponential; AVCHD may be slower, but it won’t be like making the jump from DV to HDV. Now that we know the processing load is in the ballpark, let’s shift the focus to quality.