Review: Microsoft Expression Encoder 3

All that said, the longer I work with VC-1 encoded files, the more I like H.264. Figure 3 shows you why. On the left is a VC-1 file produced by Expression Encoder 3, on the right is an H.264 file produced by the same program to the same basic encoding parameters. I can see using Silverlight for existing content of VC-1 encoded content, but going forward, H.264 should clearly be the codec of choice.

As it turns out, if you do produce H.264 video, Expression Encoder 3 is a very good encoder of choice, which was the biggest surprise of the review. The codec was developed internally by Microsoft and only supports up to the Main Profile, not the Advanced Profile as supported by all other tested encoding tools. Despite being nearly brand-new, the quality of the preserved detail was very similar to the best that MainConcept and Dicas could offer, with just a hint more noise on flat backgrounds. You can see this in my editor’s favorite screen shot, the Taimo clip, in Figure 4, which I’ve modified in an image editor to highlight irregularities in the background.

I don’t mean to overstate this deficit; it’s clearly below the threshold of commercially irrelevant, and for Microsoft to come out of the gate with such a strong H.264 codec bodes well for future releases. In the meantime, you can produce H.264 files with Expression Encoder 3 and know that your quality compares very well with the best encoding tools on the market.

One irritation is that the lowest audio bitrate available is 96Kbps, even for mono audio. I asked Microsoft about this, and I got the standard answer: "Based on user feedback, there was no demand for anything lower." I’m guessing that Microsoft asked the same users who told them that they didn’t care that Microsoft didn’t offer its own Macintosh Windows Media Player, and we know how that turned out.

More to the point, when I checked the audio data rate for the Windows Media Audio codec used by many Windows Media sites before they switched to Flash (and the inefficient MP3 codec), many were well below 96 kbps, including CBS (48Kbps); CNN, Sports Illustrated, and TODAY (32Kbps); and the Weather Channel (48Kbps). Why? Because 32Kbps–48Kbps is fine for mono WMA and is equally fine for AAC, unless you like overpaying for bandwidth. I’ve been testing H.264 encoders since they appeared, and to the best of my recollection, none prevented me from producing at my 32 kbps target for my SD test file. I suggest (if you couldn’t tell) that Microsoft should either ask a different set of users or rethink this decision internally.

OK, rant over (for some reason, inefficient audio-related controls really make my blood boil). To use Expression Encoder 3, you’ll have to navigate through the encoding options presented in Figure 5, which are pretty typical for most H.264 encoders. Again, in all cases, I would default to the Best setting for Complexity, which produced noticeably better quality than the Normal (3) default in my tests, and then adjust the other parameters as normal. Again, most presets use one-pass CBR, so change this to two-pass VBR for nonstreaming applications for optimal quality.

So basically, Expression Encoder was best in class for VC-1 and virtually indistinguishable from the best in H.264. What about other new features?

Smooth Streaming Templates
Relatively few users will ever use Expression Encoder 3’s Smooth Streaming templates, but they do provide a nice inner look at how Smooth Streaming works. You apply the template like any other, but rather than creating one set of parameters, the template creates multiple outputs, such as those shown in Figure 6.

You can modify any of the templates or click the plus sign to create another output file. With the Create Separate File Per Stream option selected, which is the default for all of the Smooth Streaming templates, Expression Encoder will create a separate file for each template. Load them onto the Internet Information Server 7.0 platform, and it takes it from there.