(Updated January 2014)
Some Technical Background
My choice of Web browser is dictated by my “wide-gamut” monitor.
What does that mean? First, a gamut is the range of colors that a monitor or printer can display. It’s also the range of colors contained in a color space. A color space is a mathematical scheme for representing colors as numbers, typically triads of eight- or sixteen-bit red, green, and blue values for each pixel in an image.
Most monitors have a fairly narrow gamut, approximating a color space called sRGB. Hewlett-Packard and Microsoft developed sRGB in 1996 as a “lowest common denominator” standard for images on the nascent World Wide Web, based on the color rendition of the CRT (picture tube) monitors available at the time. sRGB is certainly adequate for many kinds of pictures. That’s why it remains the standard for nearly all images on the Web, as well as the default color space— often the only available one— for digital cameras.
But sRGB’s narrow gamut lacks many “saturated” (pure and vivid) colors found in nature and in color photographs, which means monitors can’t display them. The closest it gets to a vivid green is distinctly yellowish. Deep blue sky can turn purple or gunmetal gray. And if your love is like a red, red rose, it might turn into a featureless red blob. The details of the petals are differentiated in colors well outside the sRGB gamut.
A wide-gamut LCD monitor can display many of the saturated colors that are outside the gamut of sRGB. But it can only do that with images that contain those colors, in a color space with a larger gamut than sRGB such as Adobe RGB or ProPhoto. It also needs color managed software, such as Photoshop, that can recognize the larger color space and translate it properly into colors on the monitor. And the monitor needs to be calibrated and profiled, to tell the color management system how to do that translation. (I discuss calibration and profiling, along with more details about color management, in my review of the huey and hueyPro, two hardware devices for calibrating and profiling monitors.) This complexity is the reason most users of digital cameras stick with sRGB.
When all those conditions are in place, a wide-gamut monitor can accurately display many more colors than a regular monitor, including saturated colors outside the sRGB gamut that some printers can put on paper. That means the petals of that red rose and individual blades of bright green grass are all visible, and the sky is the correct cobalt blue. Such a monitor lets a photographer use a larger color space than sRGB, to take better advantage of what a digital camera sensor or color film can record. (But those images need to be converted to sRGB for the Web. They’ll look dull and washed-out if they’re left in a larger color space.)
The disadvantage of a wide-gamut monitor (besides the higher cost) becomes immediately visible whenever you’re not working in Photoshop. Without color management— which, again, requires software that uses it— images take on comic-book or even psychedelic hues. Blue sky turns aquamarine, and flesh tones turn orange. Web surfing becomes something like watching an old color TV with the “tint” slightly off and the “color” knob turned up too high (if you live in North America or Japan and are old enough to remember that). Some wide-gamut monitors address this problem with a button or menu setting that restricts the gamut to sRGB. But that can cause problems with monitor calibration, especially if you forget to restore the wide gamut before a Photoshop session.
Browser Color Management
A better solution is a color-managed Web browser that can display images correctly. Fortunately, the current versions of all the major browsers support some kind of color management. Unfortunately, only one of them has a dependable complete implementation of color management that’s actually usable for wide-gamut monitors.
A fully color-managed browser needs to do two things. First, it needs to recognize images “tagged” with a profile that specifies the color space, and use that profile in conjunction with the monitor’s profile to display the colors correctly. The second thing is much more important, since practically none of the billions of images on the Web include profiles. The browser needs to assume that any image without a profile is sRGB, and then use the monitor’s profile to display the sRGB colors correctly. This is almost always the correct assumption, since nearly all images on the Web are sRGB.
The Mozilla Foundation’s Firefox is currently the only Windows browser that does both of those things correctly and reliably. Versions 9 and 10 of Microsoft’s Windows Internet Explorer, and Opera version 12.10 and later, correctly display “tagged” images with profiles. But images without profiles— that is, nearly every image on the Web— display without any color management correction.
Google’s Chrome can provide full color management. But only if you start it with a special poorly-documented command line option, from an icon that you create. And then only if the current version that Google has surreptitiously pushed to your computer hasn’t broken this feature. (That’s why only Firefox has complete and reliable color management.)
Of course, none of that really matters if you’ve got a normal monitor with a standard sRGB gamut. But if you’ve got a wide-gamut monitor and you’re surfing the Web with a browser that lacks color management, you might find yourself wondering whether someone spiked your coffee with LSD.
Opera was my favorite browser for years. I first started using it in 1997, running under OS/2’s Windows emulation. I found that the lean and clever Opera 2.12, though rather primitive by today’s standards, was much better than IBM’s native OS/2 version of Netscape. The installation file for Opera 2.12 fit on a 3.5-inch floppy disk.
Since then— it’s now at version 12— Opera Software, the Norwegian company that develops the browser, has emphasized usability and innovation. Opera was the first browser to include many features that are now ubiquitous, including multiple tabs, recall of recently closed tabs or windows, cookie management, pop-up suppression, ad blocking, password manager, toolbar customization, search engine bar, speed dial page, and page zooming. Some features remain unique to Opera, notably “mouse gestures” for page navigation that are much more useful in practice than they sound.
Opera’s user interface, including toolbars and their layout, keyboard shortcuts, and mouse gestures, can be readily customized to fit the way you prefer to interact with the browser. There are also special options for visually-impaired users: Text, images, and Flash content can be zoomed up to 1000%, a feature I find useful even though my visual impairment is fully corrected with glasses. And you can select a special high-contrast font and color scheme to further optimize text for low vision. These are only some of the reasons I find Opera more pleasant to use than other browsers.
That said, Opera has one significant problem, beyond the lack of complete color management (which, again, doesn’t matter to the majority of users whose monitors display only the sRGB gamut). That problem isn’t with Opera itself, but with its minuscule (and declining) share of worldwide browser usage. According to StatCounter statistics, Opera’s share of desktop browser usage in March 2013 was 1.17%. A year earlier, that share was 1.78%; for March 2011 it was 1.97%. Those otherwise dismal numbers might actually be an advantage if you’re concerned about security, since hackers are less likely to bother with such an insignificant browser. (Opera’s developers have always been very good about quickly issuing updates when “exploits” are reported.)
The most significant difficulty the small usage share causes is that the designers of Web sites for banks, insurance companies, and other large corporations increasingly refuse to accept Opera. Instead of a login page, you’ll get an ominous warning that you’re using an “obsolete,” “insecure,” or “unsupported” browser, and links to download Windows Internet Explorer or Firefox. Opera’s developers are aware of this problem, and have long provided options to send those sites an identification string that masquerades as one of the “supported” browsers.
Some analysts insist that this “user agent spoofing” confuses the compilation of browser usage statistics, leading to a significant undercount of Opera’s actual market share. Whether or not that’s true, the subterfuge often doesn’t fool corporate Web sites that use more sophisticated checks, and deny access anyway. And sometimes, even if the mask lets a user sneak through, important features of the Web site that rely on the impersonated browser’s implementation quirks might not work.
Opera seems to be in a lengthy death spiral, with an accelerating loss of its remaining market share as more and more Web sites refuse to allow it. Opera could end up like the Beta video cassette format, which lingered for years as a niche product after decisively losing the format war to the technically-inferior VHS. That may be why Opera Software announced in February 2013 that they were abandoning the Presto rendering engine that has long been the core of the Opera desktop browser. Version 15, released in July 2013, uses the WebKit engine that drives Google’s Chrome and Apple’s Safari. The new WebKit-based Opera will also provide the frequent “rapid releases” that Chrome and Firefox users enjoy. But Opera Software says that the “legacy” version 12 will continue to coexist with the new browser, at least for a while.
Fortunately, Opera Software derives most of its revenue from Opera Mobile, a version bundled with various mobile phone platforms. It has at least 25% of the market share for these devices. That’s quite healthy considering the popularity of Apple’s iPhone. (Apple has decreed that its Safari browser shall be the only one allowed in the iOS walled garden. Neither Opera Mobile nor Firefox are available for iPhone, iPad, or iPod Touch. Other browsers, such as Mercury— which I use and highly recommend— and the iOS version of Chrome, clothe Safari’s internals in a different— and possibly better— user interface.)
There’s also Opera Mini, a browser is cleverly designed to reduce the cost of mobile browsing. It connects to Opera’s servers, which convert the HTML pages from Web sites into a compressed proprietary format. The browser then displays the compressed pages on the mobile device. Compression can produce significant savings on data charges, making it worth using even on an iPhone. (Apple apparently allows Opera Mini in its walled garden because it’s not an HTML browser.)
I have long been one of the many users who assiduously avoid Microsoft’s Internet Explorer, which is now the second most popular desktop browser with a March 2013 StatCounter share of 29.3%. It wasn’t merely to protest the arrogance of a company that effectively monopolized the browser market for years by integrating their browser into their monopoly operating system. That near-monopoly created the computer equivalent of an agricultural monoculture, susceptible to every virus, worm, or exploit (until Microsoft got around to patching it). Using something else provided a useful measure of protection from those hazards.
Internet Explorer also has long had rendering quirks and proprietary extensions that violated standards promulgated by the World Wide Web Consortium, as well as outright bugs (such as lack of support for PNG graphics). Microsoft ignored the complaints, insisting that their market dominance entitled them to dictate the Web’s standards. It was thus incumbent on designers to build their sites around Microsoft’s browser, preferably to the exclusion of everyone else. And many designers did exactly that, because it was easier than supporting multiple browsers.
Competition from the upstart Mozilla Firefox, and later Google’s Chrome, forced Microsoft to finally update their long-stagnant Explorer. But even then, Microsoft spent quite some time playing catch-up. As a small example, if you’re visiting this site with a version of Firefox, Chrome, or Opera from 2010, the navigation buttons at the bottom of pages containing pictures will have rounded corners. That’s because those browsers have supported a new Cascading Style Sheet (CSS) option to round the corners for at least that long. But you’ll see the rounded corners in Internet Explorer only if you’re using version 9, released in March 2011, or the current version 10.
(That rounded corner is but a tiny example of the headaches Internet Explorer continually gives Web developers. Although version 9 finally included essential modern features long available in other browsers, a great many visitors to Web sites still use older versions. Developers of sophisticated interactive sites— Facebook and GMail, for example— have to develop and test two or more different versions of their sites. One is for modern browsers, including Internet Explorer 9 and 10; the other works around the absent, non-standard, or buggy implementation of necessary browser features in Internet Explorer 8 and earlier.)
Versions 9 and 10 of what Microsoft has quietly renamed Windows Internet Explorer are very decent browsers that have caught up to and are generally comparable to their competitors (although they still lack a complete tool for viewing and deleting cookies that Firefox and Opera include). But there’s no real reason to use them, except perhaps to access corporate Web sites that stubbornly refuse to allow anything other than Microsoft’s browser. There’s nothing they do that other browsers don’t do better.
If you’re still using Windows XP, you can’t use version 9 or 10. And if you have a Macintosh or a Linux box you can’t use Internet Explorer at all.
As I’ve already said, the Mozilla Foundation’s Firefox is my default browser by default, because of its complete color management. (But it does have problems with profiles that use version 4 of the ICC standard. If your monitor calibration device creates profiles in that format, the color management won’t work.) It has all the other features we’ve come to expect from a modern browser, and it works with just about every Web site out there. There is also a substantial collection of third-party extensions and add-ins, some of which are quite useful.
I still prefer Opera’s adaptable user interface. I particularly miss the ability to set the zoom/magnification for every page, instead of zooming text separately for each page. I also miss the mouse gestures. I still use Opera for Web development, where the lack of color management isn’t a problem. But for everyday Web surfing— with correct colors— I can live with Firefox. I like it better than Windows Internet Explorer, whose user interface I find very annoying.
Firefox is now the number 3 desktop browser, with a March 2013 StatCounter share of 20.87%. In an apparent attempt to retain the share it’s losing to Chrome, the Mozilla Foundation has decided that Firefox needs to look and act like the leading browser. In particular, they imitate Chrome’s minimalist user interface, with all the pull-down menus consolidated into an orange tab. (But they do provide the option of a conventional menu bar. You merely need to press the Alt key. Pressing Alt again, or clicking anywhere in a Firefox window, removes the menu bar. You can also “permanently” replace the orange tab with a menu bar by clicking the orange tab, selecting “Options,” and checking “Menu Bar.” To change back to the orange tab, select View → Toolbars, and then un-check “Menu Bar.”)
More questionable is Mozilla’s adoption of Google’s scheme of frequent releases that receive a new major version number. There were six of these “major” releases in 2011, and eight in 2012. Most of them were minor updates, which in the saner past would have at most incremented the number after the first decimal point. Mozilla’s schedule calls for a new release every six weeks. Version 10, the first release of 2012, had a mere handful of insignificant enhancements that most users would not even notice. To placate corporate IT managers who would have conniptions if they had to deploy a new browser version every six weeks, Mozilla made Firefox 10 available under an “Extended Support Release” (ESR) program. Mozilla will provide ESR users with security updates, but they’ll only get new versions every ten months, skipping six of the regular releases. Firefox 17 is the current ESR version.
The one good thing about the “rapid release” race that Mozilla and Google are running is that it encourages innovation and improvement that can benefit users of both browsers. It’s certainly better than the situation in the early 2000s, when Microsoft’s 95+% share gave them no reason to do anything more than occasionally patch their browser after a serious security flaw emerged.
Google’s Chrome is currently the most popular desktop browser, with a March 2013 StatCounter share of 38.07%. Chrome is distinguished by a user interface that’s about as lean and uncluttered as it’s possible to get. Google claims that Chrome is the fastest browser; but I don’t notice much speed difference between any of the browsers on my computer. I honestly can’t see what’s so great about Chrome; but the continually increasing StatCounter numbers suggest that Google must be doing something right.
Chrome provides full support for color management, but using that capability is inexcusably difficult. You have to start the browser from a command line that includes the poorly-documented “--enable-monitor-profile” option. The usual way to do that in Windows is to first right-click on the Chrome icon, to bring up the Properties box. Select the “Shortcut” tab, click on the “Target” field, press the “End” key, add the option to the end of this line (with two hyphens at the beginning of the string, but no quotes), and click “OK.”
Because of a persistent annoying bug in Chrome’s stealthy update process, you’ll need to create your own shortcut to start the browser with the option. When Google surreptitiously pushes an update to your computer, your “Google Chrome” icon shortcut is automatically replaced with a new one that doesn’t have the --enable-monitor-profile option. Although keeping users blissfully unaware of frequent updates is an essential aspect of the unique Chrome user experience, you’ll be tipped off to an update when color management suddenly stops working. Since Google shows no interest in fixing this bug, the only way around it is to create your own shortcut, and name it something other than “Google Chrome.”
A more serious problem is that --enable-monitor-profile stopped working in version 24. The breakage may be a side effect of Google’s attempts to improve the performance of hardware acceleration that uses the graphics chip (GPU) to speed up page rendering. Users of a Chrome discussion forum have discovered that adding “--disable-gpu” to the --enable-monitor-profile option restores the color management capability, at the cost of completely disabling hardware acceleration. “--blacklist-accelerated-compositing”, which only partially disables hardware acceleration, is a reported alternative to --disable-gpu. I have found that that either option works with version 26, but your mileage may vary.
Google seems to have fixed that problem some time after version 26. I have verified that color management works correctly without the additional flags in version 32. I also noticed that images tagged with profiles now render correctly on my wide-gamut monitor without --enable-monitor-profile, but untagged images do not. Adding the option makes untagged images render correctly. But I’m leaving the information about the workarounds, as they might be useful to try if color management breaks again.
Some people are suspicious of Chrome, believing that it surreptitiously tracks browsing and sends the information to Google so they can sell it to advertisers. But what it actually does send to Google is the same as any other browser that offers search “completion” or “suggestion” when typing into a search toolbar.
I haven’t tried Apple’s Safari, which is based on the same WebKit core as Chrome. On its native Macintosh OS X, Safari’s integrated user interface is spectacular, especially when used with Apple’s Magic Trackpad. But the Windows version never offered anything compelling, which is presumably why Apple discontinued it.