The Web Video Codec Test
The goal of this test is to compare the video quality of the most widely used web video codecs. The reason that Divx, Xvid, ON2, etc., were not tested at this time is because they do not have anywhere near the comparable internet market share of Windows Media, Quicktime, Real, and Flash. On top of that, not all codecs are designed to be used for internet video. Also note that the H.264/MPEG-4 AVC codec currently has zero market share and zero player penetration. You will have to download the beta version of Quicktime 7 just to see those clips. They were included in this test because of the incredible potential that this video format has. There was no attempt to make any comparisons between audio codecs in this test, because they use so much less bandwidth than video does. However, in a true streaming video situation, the human mind tends to notice audio skipping much more readily than things like video frames being dropped. Some media players are set up to give the audio priority over the video in a true streaming situation, for just that reason.
While the above paragraph makes the inference that Quicktime is a codec, it is in reality a player container that could, in theory, utilize just about any codec. The same thing is true of the AVI media standard. So you will be downloading several Quicktime video test clips, but each of them will be using a different video codec. The Windows Media player functions as a sort of limited player container as well, because it plays back older versions of the Windows Media format, plus a very few select things like MPEG-1, as well. The H.264/MPEG-4 AVC codec is essentially a playerless video standard, as is MPEG-2, MPEG-1, generic MPEG-4, etc. So there are both media player standards and codec standards. Some are proprietary, some are open source, but most standards have to be licensed before they can be used commercially.
Web Video Codec Evaluation
The first thing to do is to make sure that your monitor is adjusted properly, particularly in the brightness and contrast ranges, because that is what sets the black point of the monitor. If you are running an lcd monitor, you will never be able to get it set at the proper deep black level. Just get it as close as possible, per the following excellent faq link. Don't worry about the Photoshop color management for this streaming video codec testing:
http://www.aim-dtp.net/aim/photoshop/v5/ps5_rgb_calibration.htm
One of your goals is to choose between picture quality differences as a function of codec efficiency, versus picture quality differences as a function of the way that the media player affects picture quality. You should be evaluating the clips at the native 640x480 resolution, but you should also be blowing them up to full screen size as well. Defects become a lot more apparent at full screen size, but then again, how well does the media player and codec scale the picture size?
Web Video Codec Test Results
You will be evaluating nine different web video codecs, plus an MPEG-2 clip that was put there strictly for newbies to see how inefficient it is. You never want to put MPEG-1 or MPEG-2 video clips on the internet, they were not designed for that purpose, as you will see by the lousy picture quality. You will need to download the appropriate media players. As an aside, the Real player is available as a free download, just search for it. Be sure to go into the program settings and disable all of it's automatic attempts to phone home to Real.com.
Out of the nine web video codecs being tested, you will see that three of them stand out above all the rest, in terms of picture quality. Windows Media, Real, and the Nero H.264 codec will be objects of subjective evaluation. Right behind those three codecs is the Sorenson H.264 AVC offering, while everything else lags far behind that. YMMV, of course.
Here is the bottom line for the kind of streaming video encoding that this author does: Windows Media is the only realistic choice for making money with DRM. It also has excellent player penetration on the internet. It is second only to the Flash player in that regard, but not by much. The RealOne player has excellent picture quality as well, and decent player penetration, but some people do not like using it. The H.264 AVC codec is the dark horse of the bunch because of it's excellent picture quality, but there is evidence that this codec will require more computing overhead than the competition does. It will take months for H.264 AVC-capable media players to become common on the internet, but keep a close eye on it. Also watch out for H.264 AVC capability inside of the Flash rich media player. That is a very desirable combination of picture quality and player penetration.
Caveats: If you access the internet on a Macintosh computer that doesn't work well with the Windows Media 10 player, you should consider getting on a PC. As pointed out on the stats page of this website, Macs only make up 3% or less of the desktop computers on the internet. As a content creator, you need to remember that Macintosh hardware is not a factor on the internet, so putting up streaming video in just the Quicktime format is a guarantee of bad picture quality and not so great player penetration, especially with the new Quicktime 7 player. Remember, it's going to take a while for 822 million internet computers to download the Quicktime 7 media player.
Web Video Codec Test Parameters
Along with the web video clips, there is a text file listing all of the encoding parameters. The object was to encode all clips at the same video and audio bitrate, using two-pass variable bitrate encoding whenever possible. Constant bitrate encoding is necessary for true streaming video applications, but VBR will give you similar quality in a smaller file size, so it is more efficient. Single-pass encoding will give you inferior picture quality, always use two-pass encoding whenever possible.
|