Welcome to the blurry, but fast, browser…

People don’t believe me when I say Microsoft’s font rendering technology is better than Apple’s. At least they didn’t until now:

Jeff Atwood, of Coding Horror, shows the difference between IE and Safari running on Windows. Guess what, those fonts are the same on Macs too. I don’t like reading on the Mac as much as I like reading on Windows because of this.

Safari is fast, though, and has a UI that fits in with iTunes.

Oh, I’m in the Westin pool. Reading feeds in the pool is SSSOOO tough! Heheh.

Of course the blogosphere and professional journalist are going nuts about the new Safari over on TechMeme. Since they are going nuts about the pros and cons I’ll just sit back and enjoy sitting in the pool and let the pros argue it out.

119 thoughts on “Welcome to the blurry, but fast, browser…

  1. I’m in the readability camp. I think when people see the two philosophies on the same screen all becomes clear.

    I haven’t used a printer in years, so why would I care about typographic integrity?

  2. I’m in the readability camp. I think when people see the two philosophies on the same screen all becomes clear.

    I haven’t used a printer in years, so why would I care about typographic integrity?

  3. Ah, yes I did.

    And I must be in that 10 percent! I can sometimes see the subpixels on an LCD if the DPI is low enough. But I still prefer it to anti-aliased text or plain text on Windows. I prefer the extra horizontal resolution, even though it does mess with the color.

    By the way, you don’t need an LCD to get utility from ClearType.

    Without an LCD it’s just anti-aliasing and font hinting. You probably like it because of the font hinting, which makes things sharper (but distorts letter forms). Look at Jeff Atwood’s IE7 example. You can’t even tell the “j” in “Jujusoft” has a hook to it… it looks like an elongated “i.” That’s because they gave the j a hard hook instead of a bend, and its hook is blending with the underline.

    But anyway, the main point here is that it’s an Apples to Oranges (or Microsofts) comparison. They aren’t competing — they have different goals.

    Apple: typographic integrity
    Microsoft: sharpness

    They each do a great job at achieving their respective goals.

  4. Ah, yes I did.

    And I must be in that 10 percent! I can sometimes see the subpixels on an LCD if the DPI is low enough. But I still prefer it to anti-aliased text or plain text on Windows. I prefer the extra horizontal resolution, even though it does mess with the color.

    By the way, you don’t need an LCD to get utility from ClearType.

    Without an LCD it’s just anti-aliasing and font hinting. You probably like it because of the font hinting, which makes things sharper (but distorts letter forms). Look at Jeff Atwood’s IE7 example. You can’t even tell the “j” in “Jujusoft” has a hook to it… it looks like an elongated “i.” That’s because they gave the j a hard hook instead of a bend, and its hook is blending with the underline.

    But anyway, the main point here is that it’s an Apples to Oranges (or Microsofts) comparison. They aren’t competing — they have different goals.

    Apple: typographic integrity
    Microsoft: sharpness

    They each do a great job at achieving their respective goals.

  5. Mike A: if you come over my house I’ll show you Google Reader running on both browsers and Firefox blows away IE 7 in performance. Not sure what tests you’re doing to compare the two, though.

  6. Mike A: if you come over my house I’ll show you Google Reader running on both browsers and Firefox blows away IE 7 in performance. Not sure what tests you’re doing to compare the two, though.

  7. By the way, you don’t need an LCD to get utility from ClearType. I use it on my older monitors too and it looks great. It looks best on an LCD, though.

  8. Mark: you misunderstood me. I interviewed the inventor of ClearType. Video here: http://channel9.msdn.com/ShowPost.aspx?PostID=5233

    He told me that in some percentage of people that when they look at type that’s been ClearTyped that they see fuzzy rings of color or bluriness or something other than what the rest of us see. I see sharper type when I turn on ClearType. I can’t use a computer that doesn’t have it on (it’s far harder for me to read type if that’s the case).

    Watch the video, he explains how it works.

  9. By the way, you don’t need an LCD to get utility from ClearType. I use it on my older monitors too and it looks great. It looks best on an LCD, though.

  10. Mark: you misunderstood me. I interviewed the inventor of ClearType. Video here: http://channel9.msdn.com/ShowPost.aspx?PostID=5233

    He told me that in some percentage of people that when they look at type that’s been ClearTyped that they see fuzzy rings of color or bluriness or something other than what the rest of us see. I see sharper type when I turn on ClearType. I can’t use a computer that doesn’t have it on (it’s far harder for me to read type if that’s the case).

    Watch the video, he explains how it works.

  11. I’d say that this is like the 720p thread all over again, but you’re not wrong in saying that Windows’ LCD font rendering is better than OS X’s, Robert… you’re wrong in saying either is better. They’re different. They have different goals, and each achieves their goal.

    Jake: about 10% of people can’t use ClearType.

    Really? 90% of computer users in the world are running Windows and own an LCD? That seems way high. Even if your population is defined as Windows users, I doubt that 90% of Windows users have LCDs.

  12. I’d say that this is like the 720p thread all over again, but you’re not wrong in saying that Windows’ LCD font rendering is better than OS X’s, Robert… you’re wrong in saying either is better. They’re different. They have different goals, and each achieves their goal.

    Jake: about 10% of people can’t use ClearType.

    Really? 90% of computer users in the world are running Windows and own an LCD? That seems way high. Even if your population is defined as Windows users, I doubt that 90% of Windows users have LCDs.

  13. Robert,

    If it’s just a garbage collector fix that hasn’t been installed yet, then why does FF display similar performance to IE (both in the benchmarks and the actual pages I’ve looked at)? Seems like it would be a strange coincidence.

    Incidentally, for a non-fanboy discussion of the font rendering that has by far the best analysis I have yet seen, go to this link:

    http://www.joelonsoftware.com/items/2007/06/12.html

    Mike

  14. Robert,

    If it’s just a garbage collector fix that hasn’t been installed yet, then why does FF display similar performance to IE (both in the benchmarks and the actual pages I’ve looked at)? Seems like it would be a strange coincidence.

    Incidentally, for a non-fanboy discussion of the font rendering that has by far the best analysis I have yet seen, go to this link:

    http://www.joelonsoftware.com/items/2007/06/12.html

    Mike

  15. MikeA: Firefox has the better fonts on Windows too and in my tests I can’t see that much of a difference between in and Safari. Firefox is WAY faster on Google Reader, though.

    IE can fix its speed problems, though. They are due to a poor garbage collector surrounding JavaScript and is due to a known issue that they didn’t have time to fix before IE7 got released.

  16. MikeA: Firefox has the better fonts on Windows too and in my tests I can’t see that much of a difference between in and Safari. Firefox is WAY faster on Google Reader, though.

    IE can fix its speed problems, though. They are due to a poor garbage collector surrounding JavaScript and is due to a known issue that they didn’t have time to fix before IE7 got released.

  17. @Scott:

    Interestingly, the Flex performance delta is even greater: 64 fps in IE (cached) vs. 139 fps in Safari Windows (cached). Silverlight in IE gets similar performance to Flex (50-60 fps…not sure whether it can be installed in Safari Windows).

    The inescapable conclusion is that IE (and to a lesser degree FF) are just plain slow for Web 2.0 stuff. Ironically, users stuck running web apps in IE will now find themselves just like Apple Powerbook G4 users did a couple years ago – running things at 50-70% of the speed of alternative environments (e.g., Pentium M). But at least the Powerbook users (or at least the ones not in denial) could argue that they were sacrificing speed for a completely different platform environment. The IE holdouts will be reduced to arguing about how important it is to have lighter, more jaggy fonts.

  18. @Scott:

    Interestingly, the Flex performance delta is even greater: 64 fps in IE (cached) vs. 139 fps in Safari Windows (cached). Silverlight in IE gets similar performance to Flex (50-60 fps…not sure whether it can be installed in Safari Windows).

    The inescapable conclusion is that IE (and to a lesser degree FF) are just plain slow for Web 2.0 stuff. Ironically, users stuck running web apps in IE will now find themselves just like Apple Powerbook G4 users did a couple years ago – running things at 50-70% of the speed of alternative environments (e.g., Pentium M). But at least the Powerbook users (or at least the ones not in denial) could argue that they were sacrificing speed for a completely different platform environment. The IE holdouts will be reduced to arguing about how important it is to have lighter, more jaggy fonts.

  19. @Suzy:

    All three (Safari/FF/IE) are perfectly readable. That’s not the point (unless your vision impaired, in which case accessibility options should be turned on). The point is which looks most pleasing to the eye.

    Those of us who like smoothly rendered text, like on a printed page, prefer Safari’s font smoothing. But if you like thinner but more jagged text (more similar to a dot-matrix printer), then obviously you prefer IE font rendering.

    @Scott:

    On a P4/2.26 running XP, I get 64 fps in Safari Windows and 36 fps in IE 6.

  20. @Suzy:

    All three (Safari/FF/IE) are perfectly readable. That’s not the point (unless your vision impaired, in which case accessibility options should be turned on). The point is which looks most pleasing to the eye.

    Those of us who like smoothly rendered text, like on a printed page, prefer Safari’s font smoothing. But if you like thinner but more jagged text (more similar to a dot-matrix printer), then obviously you prefer IE font rendering.

    @Scott:

    On a P4/2.26 running XP, I get 64 fps in Safari Windows and 36 fps in IE 6.

  21. David Balder wrote: ” Scoble, you might want to rename the title of your blog to remove the “but fast” part: ** Wired: Benchmarks Show Safari 3 Is Slower Than IE7, Firefox **”

    First off, this isn’t really a benchmark. It’s some guy loading pages while holding a stopwatch. Nor does Google serve the same content to each browser. As such, the results cannot be used to claim Safari will always be slower on DHTML or general browsing functions.

    Jobs actually used an industry standard web benchmarking tool to come up with his stats. You can download this benchmark and verify the results yourself.

    http://tinyurl.com/2mz422

    Second, I don’t see average page loads for each browser. If the test was only run a few times per page, it’s not accounting for bandwidth variations over time.

    Third, it’s only testing a specific set of pages. As you can see, Safari was faster on one particular page. Given a larger set of pages in different applications, Safari on Windows could have come out ahead.

    For example, Safari on Windows and Mac OS X clearly out performs IE7 and Firefox on the DHTML bubble mark when running on my system.

    http://bubblemark.com/dhtml.htm

    Safari Windows: 93fps
    Safari Mac OS X: 98fps
    IE7 (Windows): 62fps
    Firefox on Windows: 64fps
    Firefox on Mac OS X: 73fps

  22. David Balder wrote: ” Scoble, you might want to rename the title of your blog to remove the “but fast” part: ** Wired: Benchmarks Show Safari 3 Is Slower Than IE7, Firefox **”

    First off, this isn’t really a benchmark. It’s some guy loading pages while holding a stopwatch. Nor does Google serve the same content to each browser. As such, the results cannot be used to claim Safari will always be slower on DHTML or general browsing functions.

    Jobs actually used an industry standard web benchmarking tool to come up with his stats. You can download this benchmark and verify the results yourself.

    http://tinyurl.com/2mz422

    Second, I don’t see average page loads for each browser. If the test was only run a few times per page, it’s not accounting for bandwidth variations over time.

    Third, it’s only testing a specific set of pages. As you can see, Safari was faster on one particular page. Given a larger set of pages in different applications, Safari on Windows could have come out ahead.

    For example, Safari on Windows and Mac OS X clearly out performs IE7 and Firefox on the DHTML bubble mark when running on my system.

    http://bubblemark.com/dhtml.htm

    Safari Windows: 93fps
    Safari Mac OS X: 98fps
    IE7 (Windows): 62fps
    Firefox on Windows: 64fps
    Firefox on Mac OS X: 73fps

  23. I wrote: “This isn’t a simply a “port” of Safari to Windows, but the same Objective-C source code used in Mac OS X. Essentially, it’s a full blown Cocoa application running on Windows. The included runtime DLLs are the fabled Yellow Box for Windows.”

    After hanging out on the WebKit IRC channel, it seems that Safari on Windows is not using the same Objective-C code as Safari on Mac OS X.

    WebCore, which is the heart of the open-source rendering engine, is written in C++ and is relatively platform independent. WebCore is shared between Safari on Windows and Mac OS X, KHTML, the Nokia S60 and Adobe’s Apollo runtime.

    However, Webkit (The higher level interface to WebCore) is platform specific. The Windows port of Webcore provides similar functionality as WebKit on Mac OS X, but is implemented in C++, instead of Objective-C, and is exposed to applications as a COM object.

    As such, no Objective-C runtime is required.

    (takes foot out of mouth)

  24. I wrote: “This isn’t a simply a “port” of Safari to Windows, but the same Objective-C source code used in Mac OS X. Essentially, it’s a full blown Cocoa application running on Windows. The included runtime DLLs are the fabled Yellow Box for Windows.”

    After hanging out on the WebKit IRC channel, it seems that Safari on Windows is not using the same Objective-C code as Safari on Mac OS X.

    WebCore, which is the heart of the open-source rendering engine, is written in C++ and is relatively platform independent. WebCore is shared between Safari on Windows and Mac OS X, KHTML, the Nokia S60 and Adobe’s Apollo runtime.

    However, Webkit (The higher level interface to WebCore) is platform specific. The Windows port of Webcore provides similar functionality as WebKit on Mac OS X, but is implemented in C++, instead of Objective-C, and is exposed to applications as a COM object.

    As such, no Objective-C runtime is required.

    (takes foot out of mouth)

  25. Well, I would expect some bugs for a while. However, it’s working great on my bigco network with tons of web aps (better than FF actually).

    Robert, saw some interesting information on the font rendering issue. Check out this article on Coding Horror that describes the difference in the Apple and MS approach, and why it shows up so badly (I actually don’t understand why Apple doesn’t do it the Windows way on Windows, but it is not mine to reason why…)

    http://www.codinghorror.com/blog/archives/000885.html

  26. Well, I would expect some bugs for a while. However, it’s working great on my bigco network with tons of web aps (better than FF actually).

    Robert, saw some interesting information on the font rendering issue. Check out this article on Coding Horror that describes the difference in the Apple and MS approach, and why it shows up so badly (I actually don’t understand why Apple doesn’t do it the Windows way on Windows, but it is not mine to reason why…)

    http://www.codinghorror.com/blog/archives/000885.html

  27. Looks like Safari 3 (on Windows) has a lot more problems than fuzzy text:

    ArsTechnica cites stability problems (well, this is a beta), UI screw-ups, fuzzy text (yet again), and numerous security vulnerabilities (found within hours):
    http://arstechnica.com/news.ars/post/20070612-afirst-look-safari-3-on-windows.html

    Oh, and they have a great comparison regarding the font rendering.
    http://arstechnica.com/news.media/font_rendering.png
    The first pic is Safari, the second is Firefox, the third is IE.
    Sorry, but anyone that thinks Safari’s text is easier to read is not being objective at all.

    And here’s another article on Safari’s security issues (note that Apple claimed “Apple engineers designed Safari to be secure from day one”, which was just more PR bull).
    http://www.betanews.com/article/Day_One_for_Safari_for_Windows_Becomes_ZeroDay_Nightmare/1181661606

  28. Looks like Safari 3 (on Windows) has a lot more problems than fuzzy text:

    ArsTechnica cites stability problems (well, this is a beta), UI screw-ups, fuzzy text (yet again), and numerous security vulnerabilities (found within hours):
    http://arstechnica.com/news.ars/post/20070612-afirst-look-safari-3-on-windows.html

    Oh, and they have a great comparison regarding the font rendering.
    http://arstechnica.com/news.media/font_rendering.png
    The first pic is Safari, the second is Firefox, the third is IE.
    Sorry, but anyone that thinks Safari’s text is easier to read is not being objective at all.

    And here’s another article on Safari’s security issues (note that Apple claimed “Apple engineers designed Safari to be secure from day one”, which was just more PR bull).
    http://www.betanews.com/article/Day_One_for_Safari_for_Windows_Becomes_ZeroDay_Nightmare/1181661606

  29. Scoble, you might want to rename the title of your blog to remove the “but fast” part:
    ** Wired: Benchmarks Show Safari 3 Is Slower Than IE7, Firefox **
    http://blog.wired.com/monkeybites/2007/06/wired_news_benc.html

    Oh well, you never could trust performace claims that come out of Steve Jobs’ mouth. For years he cited dubious (i.e. fixed/rigged) tests showing Mac Photoshop to be faster than Windows Photoshop as a way of demonstrating PPC’s superiority over Intel. It was BS all along, and now Jobs tacitly admits as much by switching to Intel. This latest Safari claim is simply the latest in Jobs’ long history of bogus performance claims.

  30. Scoble, you might want to rename the title of your blog to remove the “but fast” part:
    ** Wired: Benchmarks Show Safari 3 Is Slower Than IE7, Firefox **
    http://blog.wired.com/monkeybites/2007/06/wired_news_benc.html

    Oh well, you never could trust performace claims that come out of Steve Jobs’ mouth. For years he cited dubious (i.e. fixed/rigged) tests showing Mac Photoshop to be faster than Windows Photoshop as a way of demonstrating PPC’s superiority over Intel. It was BS all along, and now Jobs tacitly admits as much by switching to Intel. This latest Safari claim is simply the latest in Jobs’ long history of bogus performance claims.

  31. Funny they should call it clear type when it is actually just the opposite.

    the first thing I discovered about anti aliasing of fonts was that it always looks better on high density displays where the width of a line used to form the character is more than just two or three pixels.

    The second thing I learned is that when you use a high density display the need for anti aliasing in the first place is much much less. Another case where the solution to the problem is best left up to the hardware guys not the OS guys.

  32. Funny they should call it clear type when it is actually just the opposite.

    the first thing I discovered about anti aliasing of fonts was that it always looks better on high density displays where the width of a line used to form the character is more than just two or three pixels.

    The second thing I learned is that when you use a high density display the need for anti aliasing in the first place is much much less. Another case where the solution to the problem is best left up to the hardware guys not the OS guys.

  33. I like that Westin. If you’re not already staying in one, go check out the suites just next to the main Hotel tower… beautiful. We stayed there when it was only weeks old.

    The hike up and down those roads are a PITA to the pools… gotta love the little hotel golf cart/shuttles though. You’re on vacation after all…

  34. I like that Westin. If you’re not already staying in one, go check out the suites just next to the main Hotel tower… beautiful. We stayed there when it was only weeks old.

    The hike up and down those roads are a PITA to the pools… gotta love the little hotel golf cart/shuttles though. You’re on vacation after all…

  35. Finally a browser for Windows that doesn’t render text to look like it was printed on a dot-matrix printer! I can’t believe how bad text looks in IE/FF for Windows.

  36. Finally a browser for Windows that doesn’t render text to look like it was printed on a dot-matrix printer! I can’t believe how bad text looks in IE/FF for Windows.

  37. I prefer Safari’s font rendering. I find it easier on the eyes than ClearType. I have a general penchant towards generic non-filtered subpixel font rendering.

    I have TFTs on my desk and am happier with how my fonts are displayed on Solaris (that’s pretty close to Safari) than how they were shown on Windows. That includes the new C-fonts with Vista/Office.

  38. I prefer Safari’s font rendering. I find it easier on the eyes than ClearType. I have a general penchant towards generic non-filtered subpixel font rendering.

    I have TFTs on my desk and am happier with how my fonts are displayed on Solaris (that’s pretty close to Safari) than how they were shown on Windows. That includes the new C-fonts with Vista/Office.

  39. @15
    So Windows Safari is running on YellowBox. Could that be the reason for the huge RAM that Windows Safari is eating up? According to posts I’ve seen on variious message boards (like digg and osnews), Safari uses anywhere from 2 to 5 times as much memory as IE, Firefox, and Opera. Even to the point of eating up to 150M for only a few tabs.

  40. @15
    So Windows Safari is running on YellowBox. Could that be the reason for the huge RAM that Windows Safari is eating up? According to posts I’ve seen on variious message boards (like digg and osnews), Safari uses anywhere from 2 to 5 times as much memory as IE, Firefox, and Opera. Even to the point of eating up to 150M for only a few tabs.

  41. (apparently Safari for windows can’t handle international characters and blocks between the i and em html tags)

  42. Ohh… This post just reminded me of the all-time best interview on Channel9: The one with Bill Hill of the Cleartype-team. He had some very cool comments on who/what they were designing apps., fonts etc. for: The OS called Human 1.0. :-)

    Think I’m gonna watch it again now.

  43. Ohh… This post just reminded me of the all-time best interview on Channel9: The one with Bill Hill of the Cleartype-team. He had some very cool comments on who/what they were designing apps., fonts etc. for: The OS called Human 1.0. :-)

    Think I’m gonna watch it again now.

Comments are closed.