The Right Profile
Second, Microsoft also exposed a number of encoding options previously unavailable to those producing Windows Media Video files with either the Main or Advanced Profiles. They’re not particularly convenient to use at this time, but in some instances they deliver noticeably improved quality over files produced with the default parameters. I say not particularly convenient because these encoding options are only accessible via a command-line encoder or by changing the settings in your Windows Registry, which then modifies any Windows Media Video encodings performed on that computer using any encoding tool.
To be clear, once you install these new codecs, you probably won’t notice any difference in your encoding tool, except perhaps the addition of the Windows Media Video 9 Advanced Profile option as shown in Figure 1. For example, after installing the update Sorenson Squeeze showed only the Main Profile, not the Advanced Profile.
Even when your encoding tool does show the existence of the Advanced Profile, you probably won’t see the advanced options within the program interface, at least for a while. That’s because programs like Sorenson Squeeze and Rhozet Carbon Coder (see reviews, April-May pp. 78-85 and 68-70) must update their interfaces to take these into account. At NAB 2007, Microsoft announced that many developers of third-party encoding tools will update their interfaces to expose these options, but there were few, if any, demonstrated at the show.
At least in the short term, you’re going to have to roll up your sleeves to access, test, and use these new settings. Note that Ben Waggoner discussed these options in his tutorial, "Using Windows Media Registry Keys" (November 2006 Streaming Media, pp. 28–32). Another resource is the WMV9 PowerToy shown in Figure 2 (p. 78), which is available at www.citizeninsomniac.comWMV/. Designed by Alex Zambelli, who works on Microsoft’s Windows Media team, this is essentially a GUI for making the requisite registry settings. Also on that site are more links to the new command-line script and accompanying documentation.
A quick glance at the PowerToy reveals an impressive number of well-organized options, plus some helpful presets on the top left. The problem I faced in writing this article was how to efficiently identify the settings that would deliver optimum quality.
To accomplish this, I asked Waggoner to prepare a command-line script that he felt would optimize quality over a six-minute test file I had prepared for two codec studies that I had produced for StreamingMedia.com in 2006. As part of these studies, I had supplied the test file to Microsoft (as well as Apple, On2, and other companies), so Waggoner was familiar with the content. He was also in the process of encoding multiple hours of footage for Microsoft’s Silverlight announcement, and was obviously very familiar with the encoding options. Fortunately, Waggoner was kind of enough to agree to prepare the batch files that I used as described below. With batch files in hand, here were the specific questions I set out to answer. 1. Do the version 11 releases of the WMV 9 codec improve quality over the older versions. In essence, should you upgrade your encoding stations to the new codec (if you haven’t already)? 2. If you upgrade to the new codec, should you encode using the Advanced Profile? 3. Would the options accessed via the command-line script improve the quality of the WMV files over a file produced using the default parameters? In essence, should you spend the time to learn and use these options?4. Does encoding with the command-line options improve Microsoft’s performance vis à vis Flash and Apple’s H.264 codec? In essence, does quality improve sufficiently for you to consider switching to WMV if you’re not currently a user? 5. What techniques can you use to optimize the quality of your Windows Media files?
Should You Upgrade to the New Windows Media Video 9 Codec?
Here I compared a file I produced using Sorenson Squeeze back in January 2006 before installing the codec updates with one I encoded using the same tool after installing the updates. To be specific, the codec version used in the original test clip was 10.00.00.3802, and the Version 11 code was 11.0.5721.5145. (To derive this information, I analyzed the files in WMSnoop, from Sliq Media Technologies [www.sliq.com]).
In all tests reported here, I compared clips encoded at 640x480x30 fps, at a combined data rate of 500Kbps, with 468Kbps allocated to video and 32Kbps to audio. Except where indicated, I encoded all files until they matched the target data rate within 5% using two-pass constant bit rate encoding. To match the procedure used in the codec study, I encoded this test clip from the original DV source, while I encoded all other files from an intermediate file produced in Adobe After Effects using Algolith’s AlgoSuite plugin for scaling and deinterlacing (see sidebar, "How We Tested," p. 76).
In comparing the files, I noticed two main differences. First, in some high-motion sequences, the new codec was slightly less blocky than the older version (Figure 3). Second, in scenes with poorly designed backgrounds that displayed banding and similar compression artifacts, the new codec seemed less noisy. I saw no scenes where the older codec outperformed the newer codec. Don’t expect the difference to be life-changing, but if you haven’t upgraded already, you probably should now.
Note that files produced using the updated Windows Media Video 9 Main Profile codec should play on both Macs and PCs with no codec update. However, if you use the new Windows Media Audio 10 Professional audio codec, your customers will require an update, so be careful there. If forcing your viewers to download updates before viewing your new files is a concern, you probably should update a computer other than your main encoding station to produce some test files to cycle into your distribution system.