Multimedia
Blogs
Forum
Affordable HD
Whitepapers
Newsletters
Events
Advertisers
Blogcast
MillimeterMatrox CompressHD Test Drive: Mac
Aug 24, 2009 12:00 PM, By Jan Ozer
Table 1. Encoding results with Compressor and CompressHD.
Performance
I encoded three files with both CompressHD and Compressor. One was a 1-minute 1080i ProRes file to Blu-ray at 30Mbps, the next a 5.5 minute file to 640x480 at 500kbps (468kbps video/32kbps audio), and one at 720p to 1128kbps (1Mbps video/128kbps audio). I sent all Compressor encodes to an eight-instance Qmaster cluster on my 2.93GHz dual-processor, quad-core Mac Pro. Though I recommend multi-pass encoding in Compressor for optimal quality, some producers use single-pass encoding for draft work, so I included those times as well.
As you can see in the table, CompressHD was always much faster than single-pass. In the SD encode, CompressHD was about twice as fast as Compressor in single-pass encoding and was just less than four times faster in multipass mode. In HD mode, CompressHD was more than three times faster than single-pass, and about 6.5 times faster than multipass.
Figure 4. At these demanding encoding parameters, Matrox CompressHD (right) easily outperformed Apple Compressor (left).
Quality
While faster encoding is always good, quality is what truly matters at the end of the day. To assess this, I compared Matrox’s single-pass encoding quality against Apple’s multipass encoding. Here, Matrox easily outperformed Compressor in both SD and HD trials. You can see the clear difference in SD quality in Figure 4, with Compressor on the left and CompressHD on the right.
Figure 5. CompressHD (shown on the right) didn’t insert keyframes at scene changes, which created problems with this scene change.
The one fly in the ointment was poor quality at one very dramatic scene change, as shown in Figure 5, which is the first frame of a new scene. On the left, the image produced by Compressor is relatively clear; on the right, the image produced by CompressHD is distorted, and remains so for several frames. This is a particularly dramatic scene change from a Taekwondo dojo with very low levels of detail to a scene with extreme detail, and was the only one of 42 scene changes in my test clip that triggered this reaction. It’s unlikely that you’ll see this reaction in most normal scene changes, but, as mentioned, it’s a deficit that Matrox needs to address.
Figure 6. On the left, CompressHD delivered better detail with better color saturation and no fading.
In my HD trials, the quality difference between the two was less noticeable, but CompressHD produced superior saturation and contrast, without the faded look you see in the Compressor image on the right in Figure 6. I didn’t compare Blu-ray quality because at those high data rates, it’s impossible to tell the difference.
Just to save myself some hate mail, I should note that this doesn’t mean that Compressor can’t output high-quality video. Rather, it means that at extremely aggressive data rates, CompressHD and many other H.264 encoding tools will produce superior output.
Continue the discussion on “Crosstalk” the Millimeter Forum.