Tutorial: Encoding for Screencams

These configuration options aside, I did notice a significant quality disparity among the codecs when it came to screencam encoding. For example, one glance at the Compressor file convinced me that the Apple codec wasn’t the best choice, as quality was degraded at 30% above the target data rate. To verify that it was the codec and not the encoding tool, I encoded the test file in Squeeze using the Apple codec with similar results.

Note that Sorenson has switched to the MainConcept codec, but if the program is installed over a pre-version 5 installation, the Apple codec is an option as well, though clearly not one you’d like to use for screencams. If delivery data rate isn’t an issue, you can get very good quality at 500Kbps, about 2.5 times the target for my low data rate tests.

In an attempt to diagnose Compressor’s poor performance, I loaded the Compressor-encoded file into Inlet Semaphore and noticed that the data rate was almost completely flat with no variability for harder-to-compress areas. I verified that I had the right settings enabled and compressed again with identical results. In comparison, the 500Kbps file had all the normal peaks and valleys. The only conclusion to draw is that at low bitrates, the Apple codec simply isn’t optimized for screencam recordings.

While probably not codec-related, Premiere Pro also posted unacceptable results. Strangely, when producing at 15 fps, Premiere Pro dropped nearly a third of the frames in the file. I tried at 30 fps, and Premiere Pro dropped no frames, but while the still frame quality was good, moving quality was twitchy, as if short sections were played back twice in the jump from 15 to 30 fps. I tried working from a 30 fps project and various other workarounds with the same results.

That left two encoders standing: Episode and Squeeze. Here, I found that the results produced by Episode Pro’s Dicas codec were a clear step behind the MainConcept output from Sorenson Squeeze. At the target data rate, Squeeze was nearly always visually perfect while Episode showed some artifacts. More importantly, during extreme scene changes, Episode exhibited severe and highly noticeable distortion that took several moments to clear up. I tried to analyze the file in Semaphore to diagnose the problem, but Semaphore couldn’t open the file.

For the record, I should say that I tested Episode Pro 5.01 and was very impressed with the real-world video performance of the Dicas codec, which was commercially similar to video produced by the MainConcept codec for the first time. You definitely shouldn’t generalize the results of these screencams to real-world videos.

Windows Media Video
If you’re producing screencam recordings with Expression Encoder 2, your best resource is the aforementioned article by Waggoner. The careful reader will note that I borrowed much of the discussion on B-frames (and the term "pixel perfect") from his article, which I hereby acknowledge. None of the other encoders offered any of the advanced tweaking options available in Expression Encoder 2.

In theory, if you’re not tweaking, you should get the same quality from both the Main and Advanced Profile. I tried both profiles using both Episode and Squeeze and found that the Advanced Profile delivered lower quality with both programs. Overall, if you’re not tweaking, don’t use the Advanced Profile.

The only other noteworthy codec-specific option was the ability to set the number of B-frames. Interestingly, the quality leader, Squeeze, didn’t offer a B-frame control and didn’t use B-frames in the encoded file, according to Semaphore. In addition, when I used the max B-frame setting in Episode Pro (which was 4), the file didn’t play back normally; it essentially became stuck on the very first frame, and audio was distorted. When I scaled back to a B-frame setting of 2, the encoder skipped about 3 seconds worth of frames at the start but otherwise played normally.

In terms of quality, Squeeze is ahead of both Expression Encoder and Episode. This was consistent throughout the duration of the video.