Mozilla rejects Google's WebP image format

[nsane.forums]

Google built the royalty-free WebM video format with the sophisticated VP8 compression technology that it obtained in its 2009 acquisition of On2. In addition to advancing the goal of open video for the Web, the search giant also used On2 technology to build a new image format called WebP with the aim of reducing page load time by increasing the efficiency of image compression.

WebP uses some of the still-image compression techniques that VP8 relies on to compress individual video frames. The format is intended for use with lossy images as an alternative to the venerable JPEG. Google conducted a large-scale study demonstrating that WebP offers an average file size savings of 39 percent. Despite the seemingly impressive results, not everybody is convinced by Google's findings. Mozilla, which has officially refused to support the format in Firefox, has emerged as one of WebP's most prominent opponents.

Building mainstream support for a new media format is challenging, especially when the advantages are ambiguous. WebM was attractive to some browser vendors because its royalty-free license arguably solved a real-world problem. According to critics, the advantages of WebP are illusory and don't offer sufficient advantages over JPEG to justify adoption of the new format.

Aside from Google—which introduced official support for WebP in Chrome 12—Opera is the only other browser that natively supports the format. Despite the ease with which it can be supported, other browser vendors are reluctant to endorse the format and make it a permanent part of the Web landscape.

After studying quality and performance characteristics of WebP, Mozilla decided last month not to support the format. The WebP feature request in Mozilla's bug tracker was resolved with the "WONFTIX" label and a number of community-supplied patches to enable the feature in Firefox were politely rejected.

"As the WebP image format exists currently, I won't accept a patch for it. If and when that changes, I'll happily re-evaluate my decision!" wrote Mozilla developer Joe Drew in a Bugzilla comment.

Mozilla's Jeff Muizelaar offered a more detailed technical explanation about the problems with WebP in a blog post. His well-articulated critique sheds light on the problems with Google's testing methodology, lays out the weaknesses in the WebP feature set, and explains Mozilla's broader philosophical objections against indiscriminately adding new image formats to Firefox.

Muizelaar's complaints about Google's WebP testing methodology are familiar because they echo some of the concerns that were raised early on by other WebP critics like x264 developer Jason Garret-Glaser. The gist of it is that Google used peak signal-to-noise ratio (PSNR) as its basis for quality comparisons—a technical benchmark that experts say fails to account for how images are actually perceived. Another problem is that Google recompressed existing JPEG images rather than starting with uncompressed source files. Both of these factors raised doubts about the validity of Google's testing.

In response to concerns expressed by Mozilla and others, Google conducted a new study with better methodology which shows similar file size reductions as the original study at certain quality levels. The results of the new study were published last week.

WebP's lack of basic feature parity with JPEG in areas like metadata handling and ICC color profiles is identified by Muizelaar as another major problem with Google's format. It also doesn't add any important features that JPEG lacks, such as support for an alpha channel. He goes as far as using the phrase "half-baked" to describe the deficient WebP feature set.

Adopting a new image format in Web browsers is a big decision. Once a format becomes a part of the Web, it will have to be supported in perpetuity—adding overhead to the browser—even if it largely fizzles and only gains a small niche following. The chances of WebP attracting widespread use at this stage are very limited, so it seems prudent to avoid shoveling it into the browser.

Muizelaar argues that there is still room for further optimization by improving JPEG compression. The time that Google is putting into WebP, he says, would be better spent by improving JPEG encoders or contributing to existing next-generation image format efforts. One in particular that he cites as more promising than WebP is Microsoft's JPEG XR, which has a better feature set than WebP, but also suffers from a lack of obvious quality advantages.

Google's enthusiasm for WebP hasn't been dampened by Mozilla's criticism. A post published on Google's official Chromium blog last week highlights a number of quality improvements in the implementation and discusses the growing number of third-party adopters. Most significantly, Google is adding WebP support to its own Web applications—including Picasa Web Albums and GMail.

A new "fancy" upsampler that Google has integrated into the decoder implementation will reduce the pixelation of edges between colors in compressed images. The encoder has been improved with an experimental feature that it can selectively apply different compression and filtering behavior to various sections of the image in ways that will reduce the kind of "ringing" artifacts that are commonly seen in lossy images. The search giant also touts improved progressive loading support for WebP, which shipped in Chrome 12.

Despite WebP's present limitations and lack of clear competitive advantages, it seems like Google is still making meaningful progress. The WebP format isn't ready for widespread adoption today, but further optimization and perhaps a rethinking of the container format could someday make it successful.

View: Original Article