I've really enjoyed the conversations that have gone on on C9, especially when this video first posted.
So much so I've decided to have another shot at blogging. I did set up an MSN Spaces site some time ago, posted two or three entries but left it pretty much undeveloped because I had lots of other things going on
However, a few Niners have suggested they might actually visit a Bill Hill blog, so I'm resurrecting it.
Sorry for the delay in replying, I was out of the country for a while.
Yes, I've tried tinkering with the ClearType tuner tool, and each choice was no better than the first.
Hi Steve, thanks for responding.
There definitely seems to be something weird going on here. I can see weirdness even in your non-ClearType screenshot, which makes me suspicious there's a hidden factor at work.
In the black text in the non-CT example, character stems are different weights. This is normally due to a rounding error of the type that font hinting is designed to correct. You'd expect a rounding error like this if some kind of scaling's going on. It would
not be nearly as noticeable in a non-ClearType case, but ClearType would really expose it because of the sub-pixel color manipulation.
Even though you've set your monitor to native resolution, it could be something funky's going on in the hardware or driver. Or the scaling could be due to the 400% zoom you used (which is not likely because you're
seeing the issues at 100%, and 400% should just quadruple all pixels - but in an investigation like this you have to rule out all possibilities, assume nothing).
This is all just speculation until we know more - about the display, whether it's being addressed digitally or through an analog graphics port (one possible place scaling can happen), whether it has an odd pixel configuration, what graphics card and driver,
what version of Windows you're running, etc.
Be interesting to get to the bottom of this...
This isn't the place to have that kind of detailed back-and-forth technical support conversation. I've asked Greg Hitchcock, who runs the ClearType team, for an alias you can mail. I know he'll want to see unzoomed screen shots, for instance.
Once you drill down into the detailed implementation, he's the organ-grinder and I'm merely the monkey
I enjoyed hearing Bill Hill talk about the changes and advances in technology and the digital print world. It was very informative.
I would like to know though, where I can contact the Microsoft ClearType team and or the related IE team to discuss the ClearType feature.
I'm in a different mindset for ClearType. I personally hate it with a passion, because 95% of the text I read is black text, on a white background (or very close to that scenario).
I'm not sure if my eyes are better than others (I don't yet wear glasses), but even on the highest resolution screens that I use daily, with ClearType turned on, all I can see is the blurry colors (red, blue, green) that appear along side the letters on screen.
I physically have to squint my eyes so that the color "bleeds" into the black, so that my brain isn't cramping trying to un-blur the image.
Its not just on curved letters, or angled letters either. Even the lowercase letter "L" is a blur.
Are their tools to only apply ClearType to fonts over 24pt in size? or only to text that is not close to black on white?
In the mean time I've turned ClearType off, and the headaches have all gone away, but I would have to suspect by the ammount of effort that has gone into ClearType that there must be some benifit for the other scenarios, so I would like to know if I can use it there (or at least try it out)
Sorry to hear about the problems you're having.
Strange. Black text on a white background should be about the best case...
There's clearly something going on here either with your display or your vision.
First three questions to ask:
1. Is your screen RGB or BGR? There are quite a few BGR screens out there. ClearType can look horribly colorful on them.
2. Are you running at native resolution? If you're running at non-native, that can also be pretty bad. Or it may be that your display is somehow funky or faulty. Once in a while we run across a strange one.
Assuming you've done all of those and are still not happy with what you see, we know there's a small percentage of people who don't get along with ClearType at all and have to turn it off. It's to do with the balance between an individual's color perception
and their visual acuity. That's why we made the tuner - everyone's eyesight is different.
The ClearType team continues to work this to understand the issues and hopefully make further improvements. The vast majority of people find it a huge benefit. I can't bear to read type on a screen with it turned off. I know you'd expect me to say that but
it's true anyway - although I'm sure it's no consolation.
The Cleartype team would like to find out more about your issues. If you're prepared to post your email address, they'll contact you.
It took me half a day to fix the DPI issues in our WinForms framework and in our application. I've discovered that it's actually not very hard to design a new Form (window) in a flexible way: just use one or more TableLayoutPanels for
the global layout, and use AutoSize as much as possible. If you play with margins and paddings a bit, you can get some nice and clean results that scale very well.
Thanks for following up with this, Tommy!
How do you feel about writing up your experience, the issues you found, and how you solved them?
We can post a link to that here. I'm sure it would be very helpful to others, and maybe they have different experiences they could then document and share.
That way we can all learn together and help each other. We could build up a body of knowledge and "lessons learned" which could help folks make their sites and applications future-proof.
Yesterday, we discovered some of the problems with a higher DPI: applications have to be designed to work with a higher DPI. We installed one of our applications on a PC with a DPI of 120, and some of the controls in our application had
lost their text. We noticed that the windows and controls that did look good did not use absolute pixel coördinates and sizes, but used AutoSize and flexible layout (docking, TableLayoutPanel, FlowLayoutPanel).
The text scales to the new dpi - but the pane it was in has been hard-coded with a fixed pixel dimension or fixed pixel position, and text gets clipped. In the worst cases text and UI can just disappear because it's now covered by something else.
I've seen dialogs in some applications where you can't get out of the dialog, because you can't hit the "enter" or "cancel" buttons with a mouse - because you can't see them. If you know what you're doing, sometimes you can tab into them. But you NEVER want
to put a customer in that situation.
Believe me, if you think this is bad going to 120, try going to 204... OK, there aren't many displays at that res out there - but Dell's been shipping 147ppi Inspiron laptops for many years and that's a lot worse than 120 for these issues.
It's the same problem with websites.
If you have any fixed pixel dimensions in your site, it's not future-proof. If it's been designed to be, say 1000 pixels wide, then your site covers only about half of the screen on a 147ppi laptop (1920 x 1200). So someone gets to stare at a load of unused
If you've used fixed-pixel dimensions to place things, you're in trouble.
There are better ways to do it. Some sites use all the space available very well.
I'd like to see more people understanding and exposing this issue, and future-proofing thier own sites.
Until that happens, only techniques like zooming work to scale sites for high-dpi, or pixel-doubling (which only works if you have at least twice as many pixels on the display, so >192).
And that means any bit-mapped graphics get scaled and look bad because of aliasing.
How about we crank up the default dpi setting in the next version of Windows?
It's more complex than that. What about the hundreds of millions of people in other parts of the world who may have to live with resolutions ~96ppi for years to come? Another arbitrary decision would create as many problems as it solved.
Ideally, an LCD display should be running at its native resolution, and system dpi should be set to actual display dpi.
This raises a number of issues. But people are very aware of them, and I've tried to do whatever I could to help raise that awareness.
The randomness is appalling. Not to demean Shirley Chisholm, but is she really the most mportant subject in the past or present under the heading of "Politics"? The article on her is the only one under that heading so far "approved", which means the only one to
have gone through the full process.
Who made that decision? Or even worse, did no-one "decide", except the author and the people who reviewed it?
It seems even worse than the committee which started out to design a horse, and ended up with a camel. Or equipping an infinite number of monkeys with typewriters and assuming you'll get Shakespeare.
I don't buy it. How do I find the Shakespeare buried under the infinitely-large mound of other garbage generated on the way to it?
Maybe it'll get better, but it doesn't inspire confidence.
I'll stick with Britannica or Encarta, where I know someone decided what was important, subject matter-specialists were commissioned to write about it, and their work edited by professionals. People whose living depends on their performance.
Would you follow the instructions of an Internet Brain Surgeon?
Seriously though, it's great that you're a tracker.
A family of wolves and a 175-pound male cougar taught me everything I needed to know about human perception.
I live in cougar country. That 175-pound male once followed my trail back from where I'd found his family was denning to my front door. He just looked in the glass, scoped us out, stepped back and faded into the woods. "You know where I live, I know where you
I was taught wilderness and tracking skills by Jon Young, who was Tom Brown's student.
When you go for a walk where I live, you'd better be paying attention. There are often cougars around. If you're jogging - especially with your iPod - you're not paying attention.
In mountain lion world, if it runs, it's prey. They much prefer deer to people, but that doesn't do you much good if you're dead by the time he finds out you're not a deer. (Favorite kill method: a leap from the side, hit the deer in the shoulder with two front
paws, break its neck instantly. It's all about economics. So minimize the effort and avoid the risk of being injured yourself by flying hooves).
Now, one problem with mountain lions is that when young males grow up, Dad chases them off his territory. They have to find their own. Often they're very hungry at this point, and will eat anything.
They also tend to wander into suburbia, where they have to be tranquilized and released in a wilder place. I have friend who does this for Fish and Wildlife in Washington.
Once you've rediscovered your submerged wilderness perception, you see lots that other people never notice, as you say.
And until you understand how that perception works, you'll never write software that's ideally suited for humans,
Well, we have actually done the ClearType in portrait/ landscape thing, where the pixel striping is horizontal instead of vertical.
TabletPCs know about this and change when you rotate them, I believe. I'll have to try and hook Greg Hitchcock into this discussion. Greg runs the ClearType and Readability Research team and knows more of the detail of implementation than me.
There's still a lot of benefit, as you say, although vertical striping is better because we also get to use 300ppi spacing as well as fix character shapes.
The two worst things you can do to people who're trying to read:
1. Make them scroll. Designers and typographers have spent 550 years developing optimum combinations of line legnth, leading etc to avoid reading the same line twice. When you force people to scrooll, they always end up reading some lines twice which breaks
the flow of meaning. That's why the New York Times Reader is paginated - there's no scrolling.
2. Flashing things. Homo sapiens 1.0 has a visual system which makes movement a Priority0 interrupt: it takes precedence over everything else (because it's a survival mechanism). You can't stop it, it's totally automatic. So reading a website with flashing
or moving stuff is like trying to read a book in a cage full of lions. Advertisers know all about the Pri0 interrupt - that's exactly why they do it...
The Digital Declaration of Independence
We hold this truth to be self-evident: That every human has an equal and unalienable right to the means to create, distribute and consume information to realize their full potential for Life, Liberty and the Pursuit of Happiness - regardless of the
country they live in, their gender, beliefs, racial origin, language or any impairments they may have
I took a look at your blog. You've done a great job here, within the limitations of Web browsing today. Line length seems good.
Maybe this is a personal preference, but I'd try tightening the line spacing just a little with that line length. Not much - don't do anything drastic - just experiment with very small decrements on what you have. You should know when you hit the sweet spot.
The 13 point type is just a little too large for me. But people read at different distances from the screen, and they have different eyesight. Older people, or those with vision difficulties may want it even larger.
Typeface, type size, line length, leading, margins, page size etc are all inter-related variables which have to be tuned to work together.
I wrote about this in detail in my 1999 paper called "The Magic of Reading" (which I know is out there on the Web). I called the technology of type OSPREY, or Optimized Serial Pattern Recognition. Since I was out in the woods tracking a coyote when I had the
realization, I tacked "EY" onto the end of the acronym.
There's a set of OSPREY "sweet spots" for 9, 10, 11, 12 and 13 point in which all the variables are tuned to work together (There are also sweet spots for people who need larger type).
What you really want is not just to change the typesize, but change all the other variables related to it at the same time. That's what we did with MS-Reader, and that's why it's a good reading experience.
On the Web, you can make the typesize bigger in the browser by choosing "make text larger", but nothing else changes.
Anothe problem is you really want to not have as much unused space on the screen when the browser window's maximized. That's not as straightforward.
These are not criticisms of what you've done. I think you've created a pretty good workaround for today's Web. So these are general criticisms of the environment in which we all have to work today.
I'm thinking more in the area of what the Web needs to become,