Posted on April 13, 2011 at 17:00 by Kevin Chung in Technology
Recently Google announced that it will push the WebM video format by dropping support for H.264/AVC in future versions of its own Chrome web browser. Google does this in hopes to push the usage of royalty free video formats. But there’s certainly more at stake, right?
I’m a strong believer in standardizing, also when it comes to video. And the fight between video formats to become the default for HTML5’s video tag is on. Google has a great advantage in this fight: it owns the most popular video streaming site on the web, YouTube.
The eventual winner of this format standardization fight has a huge advantage through the default support in the Chrome browser. On the other hand, it’s very easy to add support for H.264/AVC or other video formats through 3rd party plugins. Also, Chrome ‘only’ has a market share of approx. 11%.
Nevertheless, with online video being one of the top pass times on the web, there’s enough at stake for the big players to all want their share of the market. And just like any other standardization war, the fight seems to be political: it’s focused on getting their video format to be the only one with default support in HTML5′s video tag.
Google is trying to make the fight more fair by using the open source approach in defining the WebM format. But just how “open” is the process? Sure, there are other parties actively involved in the development besides Google. But which ones are they? We surely seem to be hearing very little of/from them.
VP8 vs H.264/AVC
Another issue I have with the choice Google made is the codec being used. Google bought the VP8 codec from On2 Technologies and uses that as the core for coding the video stream within the WebM video standard.
I understand that owning the codec will enable Google to make it open source, but I can’t keep from wondering why Google did not take up a completely different approach and focus on enriching the end user experience using existing, proven technology. Technology like the H.264/AVC standard.
H.264/AVC is widely adopted and integrated in many video-related products for both consumers and professionals. And it’s still very capable of meeting the demands for video today: delivering high quality video, preferably in full HD resolution, at a modest to low video bitrate.
Good H.264/AVC encoders such as MainConcept and the free X264 World Class Encoder are good software examples. And from a hardware perspective, decoding H.264/AVC coded streams through the use of ASIC chips has already been widely adopted in discreet graphics cards, APU’s and standalone STB’s.
Now why should we throw away all of this adoption in favor of the WebM video format? Just so no royalties have to be paid? Such statements don’t do the business model associated with the H.264/AVC standard justice. But what’s more, the parameters used to (favorably) compare WebM to H.264/AVC are simply wrong.
As for TV broadcasting I believe H.264/AVC will eventually replace MPEG2 since the WebM format has no intention of getting into any other market than the Web.
Video quality
Metrics like peak signal-to-noise ratio (PSNR) and the structural similarity index (SSIM) are often used standards for objectively measuring video quality amongst different lossy video encoders. However, we should note such metrics don’t always fairly represent the human visual perception. Especially in more extreme cases like low bitrate encoding, there can be quite a difference between objective and subjective quality.
Therefore, I prefer taking into account subjective criteria and simply have a good look at the video. A videophile can easily tell determine the quality. It has been proven that encoders that adopt psycho-visual optimizations deliver a better end result in real life, even when PSNR and SSIM metrics say the opposite.
The VP8 video codec is still in development. Early test results show it doesn’t match the H.264/AVC quality just yet. Rumors are that VP8 shares much similarities with H.264/AVC, but that it doesn’t have the optimizations associated with it. But you never know whether those similarities won’t lead to lawsuits for patent infringements and other legal mumbo jumbo.
Green Parrot
So there are some challenges ahead for Google. But as mentioned, they have one huge advantage: owning YouTube. The world’s premier video website is certainly a great tool to drive -or should we say: enforce- usage of the WebM video format.
Google now wants to use technology obtained through their acquisition of Green Parrot Pictures to improve the visual quality of their video, especially videos shot with a cheap camera. That is: videos made by the average consumer, probably making up some 95% of all videos on the web. Google claims the improvements will reduce bandwidth and improve playback speed.
Now there are problems with these statements. What are they exactly promising here? And more importantly: is it really relevant in that they deliver benefits to the end users?
I’m not sure. One example: showing the differences for noise reduction using YouTube does not really make sense because YouTube is very limited in its bitrate (quite understandably if you consider that over 2 billion (!!!) videos need to be served every single day). The last thing you want in your source video is noise or grain, as it gets washed away anyway.
When it comes to filtering your video to improve the visual quality, this should really be done manually and can even be considered an art. No video is ever the same and automated filtering will be suboptimal.
Talking about suboptimal results, the most important determining factor is of course the initial source quality a user uploads to YouTube. Although video cameras are getting better and better many consumers simply don’t have the knowledge to minimize losses in the chain. What’s the use of filtering video that suffers from e.g. severe blocking. (I’ll cover good video quality and filtering more in-depth in an upcoming post.)
I surely hope Google won’t overdo the filtering. Consumers tend to be easily biased by huge differences like extremely abundant and saturated colors while in the eyes of a professional they are a deadly sin. If you put the WebM sticker onto it, users will believe WebM always makes video look nicer while it’s actually just only a video format.
Final thoughts
You can probably tell that I am more of an H.264/AVC supporter. But that doesn’t mean that I won’t give WebM a chance. If Google puts the focus on encoding quality and user experience, I feel it has a good chance of success.
What’s more: there’s even a thing I love VP8 for: it doesn’t support interlacing!Interlacing does not make sense today –read my previous blog post to read why.
So far, so good. Let’s hope they keep on making the right choices to make WebM the true de facto standard for web video!
Interesting infographic on the 'video war' by HTML5test.com. Click the image for large version.
