Talk:ClearType

Computing B‑class Mid‑importance

	This article is within the scope of WikiProject Computing, a collaborative effort to improve the coverage of computers, computing, and information technology on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.ComputingWikipedia:WikiProject ComputingTemplate:WikiProject ComputingComputing articles
B	This article has been rated as B-class on Wikipedia's content assessment scale.
Mid	This article has been rated as Mid-importance on the project's importance scale.

Rewrite

I've taken the liberty of largely rewriting this article and adding two new illustrations. Since the article on subpixel rendering goes into the detail on theory and math behind this technology, I modified this article to provide a description more suitable for non-specialists, such as ordinary computer users curious about ClearType and how it works. The article now points explicitly to the Subpixel rendering article for readers who want all the details on how the technology works internally.

The additional illustrations show close-ups of ClearType effects. They are not photographs, but simulations based on the screen captures that were in the previous version of the article (and are still there, merged with the new ones).

— Agateller 12:46, 21 January 2006 (UTC)[reply]

Other

Do PDA's with vertically oriented screens have vertically stacked subpixels?--Dwedit 03:15, 14 Dec 2004 (UTC)

This article should be merged (or moved rather) into the Subpixel rendering since ClearType is simply microsoft's trademarked term for subpixel rendering. 138.89.177.98

No. ClearType is a patented type of subpixel rendering that is better than the obvious method. Notice the low-pass filter specs for example. AlbertCahalan 18:04, 25 May 2005 (UTC)[reply]

Nevertheless, most of this is an entirely generic description of subpixel rendering, and that is the heading it should be under -- with perhaps a note about Microsoft's implementation, or a link to a new page here that actually explains what allegedly sets ClearType apart. Haeleth 19:29, July 21, 2005 (UTC)

I support moving it. Someone who just followed a link to this might come away thinking that Microsoft invented sub-pixel rendering. I made that mistake at first. Algr

page

I agree with last comment. Whilst Wiki should describe that Cleartype is Microsofts trademarked term for subpixel rendering, the actual explanation and definition of subpixel rendering does not belong here. Wiki doesnt seem to do this for other trademarks e.g. the page for "Pepsi" doesnt tell you how Cola is made, so why this time?

I also agree with the above comment. This page's contents should be under subpixel rendering. 18:48, 2 May 2005 (UTC)

While I can agree, I don't agree that "this page's contents should be there". Some of it may better belong there though. -- Northgrove 21:42, 23 April 2006 (UTC)[reply]

In practice

The entire 'In practice' section is horribly mistaken.

Firstly, the image used to show off subpixel rendering is not approproate - they are spheres which your brain does not need subpixel accuracy to determine. When viewing fonts, on the other hand, you can tell if something is not just right. Therefore, the image should show off text rendering. It matters greatly how closely spaced each character is from one another. With some fonts, moving a character one pixel away from another makes it appear too widely spaced. If you move it back, it then looks too close. Having 3x the horizontal resolution helps resolve this, allowing you to move the character a fraction of a pixel. Subpixel rendering allows this, and it is noticeable. Such an image would be far more valuable in determining whether or not ClearType works - especially considering ClearType technology was created for the purposes of font rendering.

Secondly, to prove his point, the author shows actual photos of his screen. But, they are at the same resolution as the original images. That is similar to showing off the resolution of a new 21" monitor by showing a picture of it on an old 13" monitor. The image needs to be zoomed in at least 3x to see if there may be 3x the resolution. Yes, I know the images link to larger images, but this does not change the fact that the images on display on the page itself have absolutely zero value. Steve Gibson of GRC.com has done the proper thing showing off his own subpixel renderer which clearly shows that it does improve the resolution considerably: http://grc.com/ctwhat.htm (view the images under the 'Through the Looking Glass' section). In short this entire section needs to be removed, since it is completely wrong. It should be replaced with the images that Steve Gibson has produced, with his permission, that proves that subpixel technology works. 18:48, 2 May 2005 (UTC)

No where on that site does Steve Gibson mention applying a low-pass filter. This is important. It prevents all sorts of nasty color fringe effects. I suspect he chose his examples carefully to avoid color fringing. AlbertCahalan 18:04, 25 May 2005 (UTC)[reply]

In http://grc.com/cttech.htm he do something end up just like using low-pass filter.

Yes, that is correct. On the main page, Steve Gibson explains it without any low-pass filter, to make it easy to understand for beginners. But, when you read on, you'll realize he uses a low pass filter. In fact, his demo program allows you to change the filter in real-time, so you can see the color fringing that occurs when it is incorrect. 15:15, 9 August 2005 (UTC)

Illustration

Using a table to illustrate the pixel pattern is inappropriate, since different browsers can display it in different ways. Can someone make a PNG graphic, or make a close-up photo of a display? —Michael Z. 2005-08-4 17:15 Z

I think a close-up of a display is a little too much, and no camera is probably able to do such a close-up. I uploaded a schematic picture, though. - Sikon 03:30, 21 August 2005 (UTC)[reply]

Good work. Thanks. [late reply; just saw this] —Michael Z. 2005-10-26 04:56 Z

Current bottom right illustration looks incorrect. The image doesn't contain any colours. Just grey smoothing.--CrazyGoldy 05:58, 2 February 2006 (UTC)[reply]

fonts?

Some fonts (eg. Comic Sans) don't have subpixel rendering like others (eg. Arial). Is it because of the font or Windows specifically disable it to that font? Also is East Asian script render with Cleartype normally possible in windows? So far I haven't seen any one that have any anti-aliasing at all.

Subpixel order

How does the underlying software know which of the three subpixel colors should be used to render on the subpixel layer. This article assumes that subpixel are in the left-to-right order "RGB".

Subpixels

In the above example, the subpixel left to a character is blue, while the subpixel right to a character is red.

What if, say, the order was "BGR" - red and blue are interchanged, therefore subpixel rendering could be wrong, leading to a rather blurry font character. Either the monitor driver tells the operating system the correct subpixel order or a industry wide standard regarding subpixel order exists. --Abdull 23:00, 10 February 2006 (UTC)[reply]

There are LCD monitors where the order is BGR, and on them the default Cleartype implementation indeed looks bad. For that purpose, Microsoft has an online Cleartype Tuner, where one can soft-tune various delicate parameters of cleartype, among which is the subpixel order:

Cleartype Tuner

Cleartype and CRT Displays

The following remark appears in the main article, under "Display requirements":

displays that have no fixed pixel positions, such as CRT displays, are incompatible with ClearType

OK, this is totally wrong. Cleartype does work on CRTs. By "work" I mean "has visible effect". While it may not do exactly what it ideally should do on LCD displays, text does look completely different - it is smoother and wider. Now, some people like this effect, others don't (claim that it looks blurry), but it is there. Exact appearance may also vary between displays, which could explain the different opinions on it.

Following are two links showing the same screenshot, taken once with "standard" smoothing and once with "cleartype". Compare it on a CRT of your choice. Screenshots were taken personally by me, so there are no copyright issues.

Standard Cleartype

I agree; ClearType on my CRT looks great, and clearly as good as on an LCD. It's among the first things I turn on on a new Windows install myself. I don't really understand why it looks so good (at least as good as regular greyscale antialiasing), but it does. -- Northgrove 21:40, 23 April 2006 (UTC)[reply]

I think you are confusing Cleartype with ordinary text anti-aliasing. Cleartype includes anti-aliasing, but on a CRT, the text would look best with anti-aliasing (Grey pixels) but not Cleartype (Color fringes). BTW, those frame grabs come from your graphics memory, so they show that Cleartype is on, but don't show what effect it has on your monitor. Algr 06:01, 28 May 2006 (UTC)[reply]

Cleartype is a set of algorithms for both subpixel rendering and anti-aliasing. When used on LCDs it combines both technologies, and when used on CRTS it uses only the latter; but even in the latter case Cleartpe it is still a specific, non-generic implementation of anti-aliasing, so it is definitely meaningful to talk about Cleartype on CRT screens. Simxp 16:24, 3 January 2007 (UTC)[reply]

Color fringing

In this magnified view, it becomes clear that, while the overall sharpness of the text seems to improve, there is some color fringing of the text. At normal viewing distances, however, only the sharpness is perceptible, and the color fringing becomes invisible.

I hope the author however realize that the point is to have this "color fringing" and NOT a greyscale antialias effect. I thought it really comes of as he/she doesn't and that it's a "problem that you don't really notice so much". -- Northgrove 21:36, 23 April 2006 (UTC)[reply]

The illustration and description of color fringing are highly misleading. The color fringing shown in the enlarged "Wikipedia" text does not exist! What this enlargement shows is erroneous color interpretation caused by the use of a screen capture or "zoom in" program that does not take individual subpixels into account. The colors shown in this enlargement do not exist in the actual displayed text; they are an artifact caused by the "zoom in" program's erroneous interpretation of the subpixels.

The following thought experiment will demonstrate this. Imagine a typical LCD display, where all the subpixels are the same size and shape, with identical borders between each of them. The subpixels are not grouped physically into "pixels". Then imagine ordinary black and white text displayed on this LCD with no anti-aliasing. It would show up as pure black and white in an enlargement. Now imagine the same text shifted to the right by one subpixel. An enlargement using a program unaware of subpixels will now show all kinds of colors. But to the human eye, the display will look identical. There will be no more visible color fringing than there was in the original display. In effect, shifting the text right has simply converted the display from an RGB subpixel display to a GBR subpixel display.

If there were actual physical boundaries between the RGB subpixel groups, separating them into discrete pixels, that would be a different story, but I've never seen an LCD display built like that.

There is color fringing that can be seen in certain cases with ClearType, but it's not the type shown by this enlargement. I will update the article when I get a chance... --Michael Geary 15:15, 21 July 2006 (UTC)[reply]

I can see the fringing - although I have to look VERY close. Algr

Yes, but any color fringing that you see is unrelated to the supposed color fringing shown in the "Wikipedia" enlargement. The colors in that enlargement are an artifact of the screen capture; they are not present on an actual physical LCD display. --Michael Geary 17:10, 30 July 2006 (UTC)[reply]

I tried making my own version and it came out exactly the same. The pic does reflect what I see on my actual LCD display. Maybe it isn't activated on your screen. Algr

ClearType is definitely enabled on my display. It's the first thing I do whenever I get a new notebook computer. :-) When you said you made "your own version", what exactly did you mean? Did you use a screen enlargement or "zoom in" program, or load a screenshot into a photo editor and enlarge it there? Any of those techniques will result in the same sampling error seen in the "Wikipedia" enlargement. Most screen enlargement programs work at the pixel level only - they do not take subpixels into account at all. But an LCD display does not have pixels. It only has subpixels, and a screen enlargement based on pixels cannot give an accurate rendition of subpixel rendering. The colors it displays are not present on the actual display, because your eye does not group the subpixels into pixels in the same way that a screen enlargement does. That is what makes subpixel rendering possible in the first place. --Michael Geary 16:33, 31 July 2006 (UTC)[reply]

I can see it every way. Screen zoom makes it most obvious, but the same fringes are there when I use a magnifying glass. I also took a shot of my screen with a digital camera, and it is there. Algr

A screen zoom doesn't make color fringing more obvious. It shows colors that aren't there. It creates fringing that doesn't exist on the actual display as seen by the human eye.

I prepared a couple of images that illustrate the point. First, the letter "W" as enlarged (misleadingly) by a program that works with whole pixels only:

Now, the same letter "W", this time enlarged with a program that correctly displays the individual subpixels as they appear on an RGB LCD:

As you can see, the two images don't even resemble each other except for the basic letter shape. The first image has tan and brown pixels to the left of each stroke, and shades of blue to the right of each stroke. These are the false colors I've been talking about. They are an artifact of the sampling error caused by the whole-pixel enlargement.

Those colors do not appear at all in the second image, and they don't appear when the normal-size W is viewed with the eye. Your eye does not lock itself to whole-pixel boundaries as a naive screen enlargement does.

I'm not saying there isn't any color fringing. What I'm saying is that a typical whole-pixel enlargement shows color fringing that doesn't exist on a real display. If you look closely at the second image, you can see where there will be some color fringing, but it is nothing like the false colors shown in the first image. Does that help clear things up? --Michael Geary 06:57, 1 August 2006 (UTC)[reply]

OK, now for some real color fringing. Here is the same letter W, rendered with no antialiasing, and enlarged to show the subpixels:

As you can see, each vertical stroke has a blue fringe on the left and a red fringe on the right. This of course is caused directly by the subpixel layout of the display. It's interesting how the fringing is more apparent in the enlargement of this non-antialiased text. --Michael Geary 09:19, 1 August 2006 (UTC)[reply]

I can see those colors it WITHOUT screen zoom. Algr

I can explain the above confusion as follows: First, it should be clarified that "antialiasing" (i.e. using varying brightnesses to smooth the edges) is a separate technique from "subpixel rendering". Image #2 above (called "W-enlargement-subpixel-rendering.png") combines both techniques, whereas Image #3 above (called "W-enlargement-subpixel-no-antialias.png") uses neither technique. Since antialiasing doesn't affect colors, Michael Geary's argument would have been less confusing if he had omitted the antialiasing from Image #2, as done in example #4 from the article:

Now, subpixel rendering DOES cause normally white or black pixels to have color artifacts (let's call this "color drift"), but this phenomenon is UNRELATED to the colored pixels seen when a screenshot is enlarged (referred to by the article as "color fringing"). Rather, this "color drift" occurs because the total count of red, green, and blue subpixels is no longer enforced to be exactly equal, which it would have been in a normal grayscale image. An extreme example would be a vertical line of width 1.3333 pixels: If it's rendered as a column of 4 subpixels RGBR, then there will be twice as many red pixels. Of course, for typical font curves, these imbalances will be very localized, and many people will not notice them at all (whereas more obsessive people can definitely see them). Similar differences of opinion arise with JPEG artifacts.

That said, I agree with Michael Geary that "color fringing" as described in the article only occurs when screenshots are enlarged. But I also agree with Algr that ClearType's "color drift" looks like dogshit from an artistic standpoint, regardless of whether it's easier for people to read. :-) Loqui 05:14, 19 November 2006 (UTC)[reply]

Yikes! I wouldn't be that harsh about it. I can usually see the color fringing, but only by leaning into the screen and looking VERY closely. It isn't something that bothers me in normal use - it seems to work well. Of course I'm on a Mac, so maybe the windows version is different. (Or gets misaligned on certain monitors.) Algr 17:51, 21 November 2006 (UTC)[reply]

ClearType on Macs

I'm surprised to read that this is a Microsoft technology. Cleartype (or some tech doing the same thing) is ubiquitous on all Macintoshes, but most windows programs still lack even basic anti-aliasing. Why is MS behind on something they invented? Do windows users all have to know to turn the thing on? Algr 06:05, 28 May 2006 (UTC)[reply]

ClearType is off by default in Windows XP. This is because the vast majority of displays that were being sold and used at the time of its release (five years ago) were CRTs. There are still a very large number of CRTs out there, but the transition to LCDs and portable computers has been ongoing for a couple of years, so in the future (ie. Windows Vista) ClearType will be enabled by default. Warrens 16:20, 28 May 2006 (UTC)[reply]

Please don't confuse ClearType and sub-pixel rendering. Sub-pixel rendering has been around for a long long time and was not invented by Microsoft. ClearType is just Microsoft's implementation of sub-pixel rendering. AlistairMcMillan 19:34, 28 May 2006 (UTC)[reply]

But sub-pixel rendering (I called it anti-aliasing.) is always a large improvement on any monitor - especially TV sets. (It's only bad for print, and it is rare to directly print screen data.) It's only ClearType that is exclusive to LCDs because on a tube, you can't know where any given pixel is relative to the color stripes. Algr 02:28, 1 June 2006 (UTC)[reply]

You say "sub-pixel rendering (I called it anti-aliasing.)" -- but sub-pixel rendering and anti-aliasing are two completely different technologies. Anti-aliasing works on all types of monitor; sub-pixel font rendering only works on LCDs. Cleartype is *both* -- when turned on on LCDs it combined sub-pixel font rendering with anti-aliasing; when turned on on CRTs it uses only anti-aliasing. Simxp 16:27, 3 January 2007 (UTC)[reply]

Algr -- The subpixel rendering & anti-aliasing implementation on Mac OS is a very different implementation to Cleartype. Mac OS discards all hinting information, in stark contrast to Cleartype which was built to take as much advantage of hinting as possible. In my opinion, Cleartype's method provides superior legibility for small font sizes, but of course, opinions vary; this guy has some high-res comparison photos a few screens down, have a look and decide for yourself. Simxp 16:46, 3 January 2007 (UTC)[reply]

Generic name?

Is there a generic name for what is done on LCDs to improve sharpness by activating individual color stripes? It's not "cleartype" or "quartz" because those names can also apply to grayscale subpixel rendering. I've made a new illustration of this, but now I don't know what to call it. BTW, it appears that plasma displays also do this with _images_ when given HDTV signals that are beyond their stated resolution. This is why I've observed that a 480p plasma display can look much sharper when given HDTV then when given DVD input. Algr 17:56, 21 July 2006 (UTC)[reply]

Added drawing. Algr

Here, it would seem that subpixel rendering is that generic name. But you seem to use this term to mean antialiasing in general - does anybody else? Anyway, antialiasing is a quite independent concept from spatially separating the colour components for rendering. I suppose one possible name for the latter is "componentwise rasterisation" or something like that.... -- Smjg 00:29, 4 April 2007 (UTC)[reply]

How ClearType works

Hmm... The "How ClearType works" section starts out OK, but the business about ClearType only working with lines less than one pixel wide is wrong, and the "Wikipedia" enlargement is misleading (as discussed above). I'll rework this section when I get a chance to do it justice, unless someone else gets to it first. --Michael Geary 09:51, 1 August 2006 (UTC)[reply]

No, I think it's a good idea to show the "Wikipedia" enlargement for two reasons: I can see the color fringing on some letters at certain sizes, and it helps the reader understand how the technology works. --Crashie 15:15, 22 March 2007 (UTC)[reply]

Tuner

Is it worthwhile mentioning the online tuner by Microsoft?
Sclozza 02:40, 8 September 2006 (UTC)[reply]

Apple Typography in See Also?

The see also section of this article has a reference to Apple typography. However, the referenced article deals with the fonts and typography of the various incarnations of the Apple logo, and the fonts on the original Macintosh.

Shouldn't we really be linking to Fonts on the Mac?

Arun Philip 14:57, 23 October 2006 (UTC)[reply]

removed image

I removed the image Image:ClearTypePixels2.jpg in the 'Illustrated comparison' section on the right because it is misleading.. It shows antialiased text as a 'simulated' version of cleartype. First of all, it should have make clear in the caption that it was a simulation, and not actual cleartype (it only said so in the text). But anyway, it is a useless and even misleading image since the point of cleartype is the improvement over antialiasing, which the image misses completely. I also moved the other image in that section. BlankAxolotl 04:40, 25 November 2006 (UTC)[reply]

Not every human has the same sight

I f.ex. seems to see more colors than "normal" people does. So when I Upgraded to Office 2007 (cleartype) and opened my outlook, I couldn't read my emails - everything seems very very blurred and moving around..strange - but calling number of my co-workers to read it, they told me it looked exceptionaly sharp and very readable - this seems to make cleartype more of a problem for some people —The preceding unsigned comment was added by 194.144.69.161 (talk) 21:12, 21 January 2007 (UTC).[reply]

Agreed. I find cleartype almost unreadable as it is all blurred and out of focus to my eyes. I know for a fact I have certain sensitivities to particular colours, it is why my glasses all come with a red tint to compensate, otherwise natural light causes me pain. I'm guessing people whose eyes lean more towards a certain end of the spectrum will be more likely to experience "foggytype". —The preceding unsigned comment was added by 82.39.188.60 (talk) 02:04, 8 February 2007 (UTC).[reply]

I do have the same problem as you guys and call it blurtype since that's the effect it has for me. The changes in Office 2007 just about made me crazy until I figured out how to turn it off, AND switch to a more legible font to boot. I'm not sure it's sensitivity to certain colors that's the problem, but what I do know is that it's particularly bad on smaller fonts. Larger ones are readable but still noticeable - in the article itself the lower text in the blown up "exerc" picture looks way blurrier than the upper one, even though the lines are smoother. I also see some mild color fringing if I have cleartype turned on (and before anyone asks, I have set the correct pixel matrix.) Anyone have any idea as to what to add/change to the main article to reflect this? 85.166.251.158 00:00, 12 May 2007 (UTC)frode[reply]

I think that everyone would see the lower "exerc" as fuzzier - It looks that way to me, and sub-pixel rendering works for my eyes. The image seems to have antialiasing that is excessive to what is really needed Is it a real image, or a simulation? It is news to me that some people react badly to cleartype. Your theory sounds reasonable, but we'd need a source to include it in the article or else it is original research. I think it's Google-time again. Algr 07:30, 20 May 2007 (UTC)[reply]

Unsubstantiated opinions on ClearType

The section containing the words "Nobody likes ClearType" and a decription of a "humanoid dog" going crazy after staring at a CRT monitor with ClearType have been deleted. Not only was the fact that "Nobody Likes ClearType" unsubstantiated, but the idea of a crazed humanoid dog(?) is patently ridiculous.