Macbook Pro, retina screens, hinting, the future

gfrederickk's picture

Hello all,

In light of yesterday's unveiling of its Macbook Pro (with Retina Display), seems like good time to restart this question:
What do you see as the future of TrueType hinting? The short-term value is obvious. I get it. I'm with you.
But when breaking out the checkbook, how do we measure the value of hinting in the context of improving rasterizers, iOS's limited appetite for hinting, screen resolution, etc.?

This is an honest question, not a statement. It's a wonderful, daunting, inspiring, confusing time of possibilities and pitfalls.
Love to read some thoughts on this, in light of Apple's news. Many thanks.

riccard0's picture

We had recently a similar discussion on the same topic. Maybe someone could retrieve the link.

Joshua Langman's picture

I was waiting for this thread to be started.

I think the biggest impact of the new hi-res screens is going to be how pixels are seen by web designers. Up till now, a pixel was a fixed measurement, equal to a point (on most screens). Whereas in print, pixels (or printed dots) have always been of different sizes and therefore irrelevant as far measuring things goes, in the world of the web, the length of something in pixels has always dictated how big it is, in terms of real-world size (inches, etc). Now, finally, that will have to change, as 220 pixels on one screen will be the same actual size as 72 pixels on another. The only difference will be that the first is a higher resolution — an attribute that now, for the first time on the web, is independent of actual size.

So I'm waiting to see a paradigm shift where web designers will have to specify the size of things in real-world terms, and let the browser sort out how big to make each pixel. In other words, the concepts of "resolution" and "size" on screen will start to be understood more like the way they always have been in print.

jasonc's picture

Since Mac's haven't used TrueType instructions in a long time, how does a new Mac impact TrueType hinting? Unless you think that the Retina display will lure Windows users. But I think a new Win OS that didn't use hints would have much more of an impact, wouldn't it?

Jason C

riccard0's picture

web designers will have to specify the size of things in real-world terms, and let the browser sort out how big to make [them]

A dreadful prospect, looking at history of browsers’ consistency…

the concepts of "resolution" and "size" on screen will start to be understood more like the way they always have been in print

To me it could be the opposite: in print you have the page, with its “real world” fixed dimensions and a somewhat fixed “resolution”. Screens will never give you neither of those.

JamesM's picture

> and let the browser sort out how big to make each pixel

I might be wrong, but I think that on the MacBook Pro with Retina display that resizing (to compensate for the higher resolution) is done by the operating system, not by individual programs like browsers.

Theunis de Jong's picture

..I think that on the MacBook Pro with Retina display that resizing (to compensate for the higher resolution) is done by the operating system ..

Correct -- up to a certain point, at least.
On an iPad 3, if you used 'regular' interface commands, dimensions in "pixels" are silently doubled. However, if your software generates a bitmap in memory and copy it "manually" on to the screen, the image will be too small.

Frustratingly, Apple still calls their old resolution "1 pixel" -- to make sense of high-resolution coordinates you are now supposed to deal with "1/2 pixels" as well!

dberlow's picture

"Since Mac's haven't used TrueType instructions in a long time..."

Oh?

gfrederickk's picture

> Since Mac's haven't used TrueType instructions in a long time...

I was waiting for David to jump in there. David -- I've read your thoughts re Macs and hinting on other Typophile posts. Very helpful.

> how does a new Mac impact TrueType hinting?

In a way, not much at all. But it's not so much about the new Mac. It's the wave other other manufacturers that will race to compete. Apple now has raised the bar on screen display. So where will that leave us?

Thanks folks for the response.

riccard0's picture

I found one, not sure if it's the one I had in mind. However I think it touches several related aspects: http://typophile.com/node/91063

gfrederickk's picture

Thanks riccard0. I appreciate it. Didn't see that one.

I searched for similar posts before adding this one. Found helpful thoughts in other comments streams, but not anything recent in this context. I'll give that thread a read.

gfrederickk's picture

> I think the biggest impact of the new hi-res screens is going to be how pixels are seen by web designers.

Joshua, thank you. This is another thought that keeps me up at night (in a good way). What is the impact of responsive design and relative type sizing on hinting? I don't know enough about the mechanics of raterizers and the way they trigger font meta data. But seems to me the scenario would be the opposite of the way you describe. No? In responsive design, there is no real-world fixed size. Everything is relative.

Nick Shinn's picture

In responsive design, there is no real-world fixed size.

Real-world?
It’s a veil of illusion (whatever the resolution of that veil).

quadibloc's picture

My web site makes extensive use of line diagrams prepared in a paint program, for example on this page. They would not be legible if displayed at an arbitrary resolution where 1 pixel in the image translated to 1 1/8 pixels on the screen.

Therefore, specifying image sizes in inches, for example, is not an option, since then, although the images might be legible enough on a very high-resolution display like the Retina display, they wouldn't be legible on different resolutions (60 dpi - old Macintosh, 72 dpi - common Windows, 96 dpi, 144 dpi, alternate Windows) on conventional computers.

Instead, Apple's current solution, of having a real pixel be equal to 1/2 or 1/4 of a virtual pixel, does work.

Ideally, in future, I would have to change how I specify image sizes, but how it would be done would be to specify image sizes in pixels, but then also note a base screen resolution: above twice that resolution, each pixel would be replaced by an integer number of pixels.

i.e., if I specify the base screen resolution as 72 dpi, then at resolutions from 144 dpi to 215 dpi, each pixel becomes a 2 x 2 square, at resolutions from 216 dpi to 287 dpi, each pixel becomes a 3 x 3 square, and so on.

That would combine resolution-independence with the ability to control the relationship of image pixels to screen pixels.

Naturally, for things like photos, it should also be possible to specify image size in inches, where resizing will not injure visibility.

jasonc's picture


"Since Mac's haven't used TrueType instructions in a long time..."

Oh?

-----------------
Oops, David caught me. But the effect is virtually the same.
Jason C

dberlow's picture

Lol:) I'm happy I don't have to go to déjà vu U. For me as a user, this is not an interesting Apple product. I'm sure I'll get one the next time I need a laptop, but that's just a matter of course. As a developer, the big web resometer in the sky still reads 16.6 ppm for text. So I have to break out the checkbook. What check are you asking about writing to whom?

Frode Bo Helland's picture

True, true, DB. Still, if anything/anyone is going to make hi-res displays commonplace, this is the event. Hooray! Now, those of you who don’t think hinting is worth it just gotta sit tight for a decade or two.

aluminum's picture

There is some confusion here. At least with iOS and OSX, there is no 'thinking different' (pun not intended) about pixels for web designers.

Designing something 10px square in mobile Safari, and setting type to 8px will result in the same physical sized elements on a retina display as on a non-retina display.

As previously stated, this is because apple now treats a 'pixel' as a more virtual concept. On retina displays, it uses 4 device pixels to render the one 'code' pixel.

So, generically speaking, a web designer has to do nothing to accommodate retina displays.

If they do want to give retina users higher resolution images (Again, displayed at the same size) there are a few steps they need to take care of, but it's nothing daunting.

What is frustrating for web/mobile designers, however, are the other high-density screens being developed that don't do what Apple does. Instead they just use smaller pixels. Some of the BlackBerries do this and it causes issues in that the designer sees everything smaller on the device compared to the iPhone so insists that we resize everything for the BlackBerry. However, they often fail to realize that EVERYTHING is smaller on that user's BlackBerry so the user has already likely taken steps to change the defaults to their liking to begin with.

The bottom line is that high density screens are going to cause weird problems going forward--at least until all OSes start using resolution independent rendering. But Apple is the one place where we shouldn't see too many problems.

Theunis de Jong's picture

But Apple is the one place where we shouldn't see too many problems.

Personal experience from opening a couple of freshly-made ePubs on an iPad 3 shows that is more of a hopeful statement than fact...

Ryan Maelhorn's picture

I find it ironic this question popped up on a type design board. Seems like out of all the visual digital artists out there type designers will have to change the least. Everything we do is already designed to look good at any resolution.

The first big impact this tech will have will be on computer memory, both hard drive and RAM. I've already heard grumblings from other designers about how insane the project file size gets. Can you imagine how much RAM and hard drive space you would have to have to edit a full length film? With dozens of hours of raw footage? And the after effects guys, they might as well hit render and go on a four week vacation... But of course, technology will improve, and things will get faster and cheaper.

aluminum's picture

@Theunis...what are you seeing?

@Ryan...I agree that this should have little to no effect for type design. In terms of hinting, it makes it less important.

It shouldn't affect video editing at all, though. Unless people decide to start creating film at 4x HD resolution...which they may. ;)

gfrederickk's picture

> What check are you asking about writing to whom?

Thanks David. I'm referring to a check written to a talented type technician to hint an non-TrueType hinted font. That's a sizable expense. I'd hint my alpha-bits if I could, but increasing screen resolution makes me wonder how sharper screens will affect the value of intense hinting and the ability to justify the investment.

gfrederickk's picture

> I find it ironic this question popped up on a type design board. Seems like out of all the visual digital artists out there type designers will have to change the least.

I'm not a type designer. I'm a type user. The fact is, outside of the type designer/provider universe, there aren't many people who can intelligently weigh in on hinting conversations. So I value the feedback I get here.

Screen resolution may not be a strong consideration when designing your typeface, but isn't it something you consider when making the decision to invest lots of time and resources at your foundry to add hinting information? If better screens improve the display of the typeface, that would seem to lessen the motivation of designers and publishers to invest in hinting. Wouldn't that affect your decisions?

Likewise, there are third parties who can do this work well (WebType, Typekit, etc). I imagine the service they provide would lessen a designer's urge to invest heavily in hinting.

Ryan Maelhorn's picture

Well its seems on the face of it that if the resolution goes up, the hinting should become easier to do. Eventually, perhaps, when we hit 24 000 x 24 000 pixels it should become unnecessary altogether.

But I've been wrong many times before, and will be wrong many more times in the future.

(note: various figures on the web point to human sight having a resolution of 576 megapixels, which works out to 603 979 776 pixels, the square root of which is 24 576.)

JamesM's picture

> I've already heard grumblings from other designers
> about how insane the project file size gets

True, but increasing file sizes has been the trend for decades. I remember the early days when I could back up my computer on a stack of floppies. Today that same stack probably wouldn't have the capacity to hold a single big video file.

Theunis de Jong's picture

Interesting stuff, all of it.

Yes, high resolution screens will, or just might, make low-level hinting obsolete. But not every screen you see around you is, or will be in thefuture, a high-rez screen. There will always be a market for properly hinted fonts, I think.

Perhaps it'll make a nice niche market, one day, when there are just a few experts left that can make fonts good looking on any screen.

As for iPad 2 vs. 3 differences: images that were scaled in e-pubs, to either a certain scale of the total width, or a certain size in pixels, opened up on the '3 at ... half their size! In Real World measurements, of course. Another thing I noticed was that same text in (supposedly) same size spaces just a tiny bit differently on the '3. Tiny bits add up, though, on an entire line. So it seems subpixel alignment was not translated into half-subpixel alignment (or double-subpixel, perhaps).

oldnick's picture

True, but increasing file sizes has been the trend for decades. I remember the early days when I could back up my computer on a stack of floppies. Today that same stack probably wouldn't have the capacity to hold a single big video file.

Making bandwidth hogs of us all, for which we pay and increasingly dear price...

aluminum's picture

Theunis...that's interesting (and obviously annoying!). I haven't done much with ePub. That certainly sounds like a bug/oversight (either on Apple's end or the ePub author's end).

Ryan Maelhorn's picture


Yes, high resolution screens will, or just might, make low-level hinting obsolete. But not every screen you see around you is, or will be in thefuture, a high-rez screen. There will always be a market for properly hinted fonts, I think.

I'm sorry but you're way off base here. Technologies judged as inferior by the masses become obsolete quickly.

As soon as the market recognizes a truly superior and readily available technology, all the money will switch over to that. Nobody will care about your 15 year old mint condition monitor. If the new fonts don't work on your system, the response from Support will be, "oh well, buy a newer computer." This is just the reality of markets.

Nick Shinn's picture

Nice to see my fonts improving with age :-)

Té Rowan's picture

@Ryan – The only choice the consumers get is that which the manufacturers afford them. And more often than not, it is marketing that sways a consumer's mind. Had Sony gone with more aggressive marketing than they did (including licensing to other manufacturers), JVC's VHS might well have become the also-ran instead.

joeclark's picture

I trust you are all aware that pixels are a strictly notional unit in CSS? A pixel positively does not correspond to one individual dot. (In fact, rather akin to the reference kilogram, you have to avail yourself of a reference pixel.)

I make this point because there is actually a great deal of experience in mapping presumed-immutable physical dimensions to virtual ones. (Have you ever drawn a dimensioned architectural drawing in AutoCAD? Is the drawing the same size as the building?)

Té Rowan's picture

Not everyone knows that W3's world is 96dpi through and through.

Theunis de Jong's picture

Joe, when I right-click your avatar image and choose "Properties", I get a value of "60 x 44 pixels". That's the same as when I make a screenshot and measure it in Photoshop:

So what values do you see? What measurements "is" the actual image, in "pixels-in-single-units-of-one-RGB-color" when you download the file and inspect it with a binary editor? Surely that would circumvent any 'file to screen' translation?

dberlow's picture

" (Have you ever drawn a dimensioned architectural drawing in AutoCAD?"

100s, only you don't need to dimension them anymore, you output them with dimensions if you want. All the objects know their "actual size".

"Is the drawing the same size as the building?"

Ideally, exactly the same, in CAD. Architectural scales are for the analog world. And if you're CADing anything with type on it, like signage or emergency fire systems, and you want to know its actual size, you better be drawing at 1:1 or you get hideously unuseful font sizes.

Whoever gave the web the ideas it has about sizing, IMHO, should be rendered.

aluminum's picture

Theunis: OSX/iOS treats 'pixels' as virtual units.

Or, rather, Apple calls them points. I think. I dunno. The terminology is weird.

But the point is that on a non-retina display, 1 pixel measurement = 1 pixel on the screen. On a retina display, 1 pixel = 4 pixels.

That means that for most software, nothing changes at all on the retina display. Your UI will still be the same size. However, elements of said UI will look sharper if the software accommodates that.

Té Rowan:

The W3C, being about the web, really has no concept of DPI.

Té Rowan's picture

OKAY, PIXELS per inch, ya ferschlutiweeblefetxcer pedant!

paragraph's picture

If I remember correctly, the square pixel (1 px = 1 pt, screen resolution 72 ppi) was a technological breakthrough by Apple, in an era when computer graphic screens used a rectangular pixel (I had Ventura Publisher on Hercules? GEM? or VGA?). As a result, the Mac was the only computer that could display type at anything like its real size.

joeclark's picture

A day without a cranky, supercilious rejoinder from David “Fools!” Berlow is as a day without sunshine.

The fact remains that the pixel is a notional unit, not an immutable object.

Also, David, answer my mail about caption fonts.

quadibloc's picture

A pixel had better correspond to one individual dot, if that dot is of a reasonable size (somewhere between 60 dpi to 96 dpi) - and, failing that, it needs to what Apple did, and correspond to an integer number of dots in each direction, and be square.

Then existing web pages will render legibly.

But ideally, a pixel should be an individual dot, and designers should have tools in HTML to cope in an accurate fashion with pixels that have unusually small dimensions. For the term "pixel" to mean one thing in a hardware manual, and something else in CSS, can only lead to confusion.

aluminum's picture

Well, you wouldn't be the first to question some of the terminology adopted by the W3C for their CSS specifications. ;)

Té's 96ppi comment *is* related to the spec. The W3C refers to a 'reference pixel' in the latest working draft:

http://www.w3.org/TR/css3-values/
"Note that this definition of the pixel unit and the physical units differs from previous versions of CSS. In particular, in previous versions of CSS the pixel unit and the physical units were not related by a fixed ratio: the physical units were always tied to their physical measurements while the pixel unit would vary to most closely match the reference pixel. (This change was made because too much existing content relies on the assumption of 96dpi, and breaking that assumption breaks the content.) "

Granted, it's still up to browsers to decide what to do and not all care. For instance, there are several high-density BlackBerry devices where the pixels are simply smaller. Which is unlike the iOS reitina implementation where the 'real' pixels are much smaller, but the 'virtual' pixels remain the same offering a more predictable rendering in terms of physical size.

At some point the operating systems/OSes will have to just do away with the concept of pixels altogether (at least from the perspective of the content authors). As we approach much higher density screens, adopting a resolution independent system would make more sense. That way 1" = 1" regardless of the particular screen. Fingers crossed.

Mark Simonson's picture

A pixel on a web page has been a slippery concept on all iOS devices going back to the first iPhone. Web pixels ("px") almost never correspond to device pixels. The page is scaled to the device based on the nominal width of the page. Sites designed for narrower widths will appear bigger than sites designed for wider widths. All images and text are scaled accordingly. (This can be annoying when you are trying to look at screenshots posted on a website.)

The only way to force exact pixel correspondence in Mobile Safari is to set the page width to the same width as the pixel width of the view (vertical or horizontal orientation) on the iPhone or iPad, as is done with web apps.

dberlow's picture

"At some point the operating systems/OSes will have to just do away with the concept of pixels altogether (at least from the perspective of the content authors). As we approach much higher density screens..."

Just for the sake of confusion, if they ever get to the "some point" mentioned above, let's call it "PostScript".

abattis's picture

high resolution screens will, or just might, make low-level hinting obsolete

I don't think so. Low level hinting is already obsolete without the high resolution hardware, because the latest Apple, Microsoft and Libre font rendering system software can render unhinted fonts well enough.

The issue is that the old Microsoft rendering system software ("GDI") is hard-coded by many Windows applications - so the new DirectWrite system software isn't used even if it is available (c.f. Google Chrome today)

I expect plugging in an Apple Retina screen to a Windows 8 computer that is running such 'legacy' software is unlikely to render text any better than the previous screen. The Retina display drivers will use 4 or 8 hardware pixels to represent 1 pixel in legacy software, so it will look the same as it always has.

Because that is what businesses want; change on their schedule, which might be never.

When you peek at the computer on the checkout stand at the grocery store, you'll often see a Windows 7 computer running a VT100 terminal emulator with a UI that looks like this:

If the software using GDI rendering is updated to use DirectWrite, then even on old screens an unhinted font will render fine.

But when breaking out the checkbook, how do we measure the value of hinting in the context of improving rasterizers, iOS's limited appetite for hinting, screen resolution, etc.?

If you are going to pay for hinting, my suggestion is to make a contribution to the http://freetype.org/ttfautohint/ project and use it to bridge the gap between now and the time when most people are no longer using applications linked to GDI.

gfrederickk's picture

This is an education re the nature of pixels. Comparing the relationship to CAD helps (I'm a Lightwave user). I see a dimly lit rabbit in the distance, beckoning me to follow deeper into the rabbit hole.

joeclark's picture

To respond to Theunis’s question: Your example contains its own answer. The image has a claimed pixel dimension but can be viewed at other dimensions. In fact, you already are doing so in the example screenshot.

The Statue of Liberty is available in only one size. The same does not hold for a pixel.

dberlow's picture

"If the software using GDI rendering is updated to use DirectWrite, then even on old screens an unhinted font will render fine."

Pass the mushrooms please.

Theunis de Jong's picture

Um, Joe, you are correct. I think the solution to the problem at hand finally dripped down into my skull (akin to Chinese water torture).

W3C indeed uses "the pixel" as a valid measurement unit, so you can ask for an image to be sized 44 x 30 pixels (either in HTML or through CSS). If the pixel were not a valid measurement unit, this would not be possible!

However -- dropping upon me like a bone thrown into the air by a careless simian --, it depends on the browser how this value gets interpreted. So in my example screen shot, using Internet Explorer on Windows, this only tells us that Windows treats a given pixel unit as ... real-world pixels. It does NOT tell us say anything for my current reading platform (Safari per iPad 2) and it won't tell us anything for the same on an iPad 3. Wait -- I feel an experiment coming up. ... Okay, when I go from landscape to portrait mode, your avatar scales down a bit.

Resuming? When we are talking about pixels, do we henceforth have to refer to "W3C pixels" (as shorthand for "1/96th of an inch", just like the nominal point is "1/72th of an inch") and "device pixels"?

Nick Shinn's picture

Joe, the Statue of Liberty comes in several sizes, the big one being impractical for souvenirs.

Rosalind Kraus wrote an interesting essay on this sort of thing, about Rodin's work, the factory he contracted to reproduce it, and the issue of copyright and originality in art.

Té Rowan's picture

@abattis – The local grocer's checkout is DOS-based and with a GUI. Ain't seen a Win7 checkout yet, but the video rental has some Linux boxes with touch screens for renters to view the catalogue on.

Richard Fink's picture

I agree totally with David, Joe, Nick, Dave (abattis), Mark, Té and Theunis on all this.

Oh, and BTW, I think the idea of pixels being "actual" and corresponding exactly to each and every illuminated dot on the screen took hold mostly because the unit of measurement is named the same as the illuminated dot but also partially from a misunderstanding about "relative" units versus "absolute" units in the Cascading Style Sheets spec.
Yes, the pixel unit is a "relative" unit, but not in the same way as an em or percent. The pixel is relative to the size and resolution of the physical screen, whereas the em or percent is relative to the computed size of the parent element within the HTML.

Write HTML pages, post HTML pages, view HTML pages. Adjust. Repeat.

Syndicate content Syndicate content