Looking at StatCounter stats, it occurred to me that they might not match the common narrative about H.264 market share. I decide to run some numbers using StatCounter stats. It turns out that during the first two weeks of 2012, on desktop, the usage-share of browsers that support WebM in HTML5 video exceeded the usage share of browsers that support H.264 in HTML5 video—except in North America and Oceania.
The total desktop bucket adds up to 97.7% globally. In other words, the non-desktop browsers that StatCounter reports in its desktop stats account for less than 3% of usage. I divided the numbers in the other buckets by the total desktop bucket to scale the other numbers to be percentages of desktop only.
The script (which you can audit) has baked-in knowledge of which browser versions support what. Only out-of-the-box support is considered, so IE9 is considered to support H.264 but not WebM even though WebM support can be added as a separate installation. As an exception, Safari is counted as if Safari for Windows always also had QuickTime installed (because the data is not granular enough to do otherwise).
The percentage means the Web usage share (as measured by StatCounter) of desktop browsers that support a given video format in HTML5
<video>. The percentages do not add up to 100%, because some browsers support multiple video formats. The numbers for Theora are always higher than the numbers for WebM, because all browsers that support WebM also support Theora but there are browsers that support Theora but not WebM.
|World||Africa||Asia||Europe||North America||Oceania||South America|
Answers to questions from Twitter.
I tried to make the point that the share of tablets is tiny relative to desktops. If Safari on iPad and the Android default browser are both counted as H.264 and no attempt is made to count any versions of the Android default browser as WebM (although some versions support WebM), the differences in world-wide numbers are lost in rounding and do not change the rounded percentages.
The desktop+tablet numbers that StatCounter provides are not commeasurable with the phone numbers they provide.
If pluggable codecs were considered, the WebM and Theora numbers would be higher, since IE9 supports WebM if WebM support is installed to Media Foundation and Safari (on Mac at least) supports WebM and Theora if WebM or Theora support is installed for QuickTime.
The H.264 numbers would not be higher, because Firefox and current versions of Opera do not support H.264 as a pluggable codec and Chromium usage is very low. (Earlier versions of Opera for Linux and FreeBSD supported pluggable H.264, but those versions have already faded out of significance.) The Windows Media Player extension that Wikipedia counts as adding H.264 support to Firefox does not actually add H.264 support to the HTML5
video element. Instead, it replaces the
video element with a plug-in instance that does not support the HTML5 video API, so the HTML5
video element is no longer there. (Feel free to try to fix the wikitruth. I have given up.)
I have no plans to do that at this time. The script is available under a Free Software license, so feel free to run it over data from a different time period.
In January 2011, they announced that H.264 support would be dropped. As of January 2012, Chrome has not yet dropped support for H.264.
Please see the source of the script.