Just installed Cubase 5.5 (generously supplied via mail by the Steinberg team because some old projects weren’t opening in pro 8).
This is almost unbelievable. Apart from some GUI elements that are drawn, like the R/W/S/M buttons, everything else looks perfect on my retina MBP. All text renders great, while it doesn’t in pro8.
I can’t believe this.
Screens coming on as soon as I get a stable internet connection.
Older versions of Cubase were part of the “pixel perfect and crisp font era.” It’s something that always drew me to Cubase.
But, it’s not that simple.
They’ve had to give that up in pursuit of the resolution-independence requirement that comes with the modern, high density display trend, and ironically, found themselves in a sort of technological “valley” where their new text engine does neither 72 DPI displays, nor high density, well.
I suspect they’ll improve it, but it’s not an easy problem to solve.
The only reason you’re seeing crisp fonts is because Apple chose a mathematically simple “doubling” of the screen you’re looking on (probably); an “@2x” or “2.0” pixel ratio, as it’s called. They did this to help bridge the gap until everything is high-density. Apple being Apple, they’ve already started prematurely to say “okay, times up…that time is NOW,” which means they’re starting to release devices with non-doubled pixel ratios.
A non-doubled pixel ratio of, say, 1.5 or 3.0 would not make those Cubase 5 style “crisp fonts” look as crisp, and those wonky ratios are unfortunately becoming more common.
Worse, Apple’s competitors don’t care about this attention to detail and have started the non-doubled high density display trend even earlier (forcing Apple to follow suit).
So, Steinberg can’t simply continue to use pixel-crisp fonts and hope all screens will be at a no-brainer 2.0 / 4.0 / 8.0 pixel ratio, where everything would remain pixel-perfect and crisp. They have to use techniques that work for 3.0 (and other “non doubled”) pixel ratios, too.
The tradeoff is dynamic scaling vs hand-edited fonts. Those Cubase 5 fonts were pixel-perfect, fixed and hand-edited by a designer targeting 72 DPI screens only.
It’s a difficult computer problem to solve and will take years to perfect. Much of which depends on Microsoft and others to supply the solutions for.
In the meantime, we’ll have mediocre font rendering.
I really do feel Steinberg is making a pretty good effort here, given the Herculean challenge.
That said, I think there are some things they could do to help bridge the gap better. I suspect that some of those older “Cubase 5 font gurus” are either gone or somewhat out of the loop now.
They need a new font guru that gets the whole “pixel perfect fonts” thing and tries to find ways to realize that as closely as possible in this new resolution-independent font engine era.
If there is such a person, and this is their best effort, then my apologies.
Btw, the test of such a person on the Cubase design/development team is simple: do you have a pixel zoom tool app that you are using when evaluating the font size breakpoints logic (and the magnifier app doesn’t bilinear filter the output, but shows each raw pixel unchanged, not “smoothed” and not anti-aliased)?
If yes, you pass the test. You’re the correct person.
If no, then sorry, you don’t “get it.” You need to step up your game!
Application and OS fonts need this kind of attention to detail.
Yeah, I’m an old school pixel-perfect ad agency guy / designer and have been following this topic closely.
I’m mostly talking about any in-app fonts. Mostly those that Cubase would be “in control of” (maybe not certain OS-level ones, so as not to break guidelines and conventions).
As for Notation, that’s going to be 100% “roll your own” from a development standpoint. Which could be good because it allows fine-grain control, but I suspect the opposite is true and it would suffer the most disregard to “pixel perfect” due to the complex nature of the glyph shapes involved.
As for compensating, yes, if all high-density displays on all Mac and PC devices were simple pixel doubling (2, 4, 8, etc.) then it’s easy. Apple started this trend with an encouraging intro: The first Retina iPad was perfectly pixel doubled. Therefore, a simple doubling of assets kept things crisp and 1-to-1. Unfortunately, Microsoft, Samsung and Google, feeling the pressure to answer Apple’s volley, had to act quickly and so they used displays already developed that didn’t well-consider these details. So, we saw a ton of non-doubled 1.5, 2.5 and 3.0, etc.
I you look at this chart under the “pixel ratio” column, you can see the whose to blame for this travesty.
1, 2, 4, 8 are all “good.” Anything other values are creating the problem.
Samsung, Sony, Microsoft and Google are to blame.
You’ll notice it was only the iPhone 6 Plus were Apple was sorta forced to do that “3.0” and break their good track record. Tim Cook was forced to deliver an answer to the market and I totally get why he had to do 3.0. It’s unfortunate, but it’s done. We’ll never see a pixel-perfect era again.
The only way we might, is if developers like Steinberg, device manufactures and OS developers collaborate on a solution. This is unlikely to happen because once you get past about 300 dpi, you reach diminishing returns with the whole pixel-perfect thing.
Unfortunately, Apple has already decided on the math that defines what pixel-perfect is and is driving the market. And the math is not quite forgiving enough to satisfy pixel-perfect aficionados (it’s a thing, a circle of designers I’m part of).
That “retina” math, and the artificial constraint Apple put on it, is ironically fighting against the good intentions Apple started with. While others are pushing the DPI number UP, Apple is keeping it DOWN (to meet bare minimum “Retina” math).
The need for 3d graphical performance on mobile is working to keep Apple’s low number for Retina the winning solution. It’s gonna be this way for a long time. Probably another 5 to 10 years.
So, 300 - 400 DPI it will be for a while (and the lack of true “crispness” that pixel fonts give us – as enjoyed in Cubase 5.)
Once you get past about 400-500 DPI, even non-crisp fonts can look crisp (the good thing we want). So, it will be a race between the “enemy” Samsung and others to force Apple (on raw spec. competition) to up their DPI.
Basically, a battle between raw resolution and raw 3d FPS performance for video games.
I honestly would not like to place on bet on this. Do games win, or does DPI importance to the average Joe, win. Probably games… sigh.
The whole ‘retina’ scam totally ignored that in print 600dpi looks better than 300dpi for text, lines and any other scenario where there are abrupt colour changes. That 300dpi is easily ‘retina’ to begin with.
Scalable fonts require extra resolution to have pixels that are ‘in-between’ colours to fill in oblique lines, so that when viewed from a distance where the individual pixels CANNOT possibly be resolved, the lines look smooth.
In other words, it is NOT enough to be JUST out of pixel resolution, but substantially better than it.
I had a Note 2, which at 1280x800 was ‘retina’ at the distances I viewed it, but the Note 3, at 1920x1080 looked noticeably better, especially for small text.
The real problem is that the screens that most people look at programs like Cubase on, are nowhere near a high enough resolution to be able to show small scalable fonts properly. They are OK for pixel optimised fonts, but run an LCD screen at anything other than their native resolution, and those look terrible as well.
Screens that are just making ‘retina’ are good at showing things like pictures, where there are not abrupt colour changes from pixel to pixel, so linear interpolative re-rendering has enough pixels over which to work, but dramatically fall down with lines and text which do, so a totally different re-rendering mechanism (the extra partial fill pixels above) is required.
So very true, Patanjali (on every point you made).
I mean, hold a piece of imagesetter film output at 2400 dpi, with some line art and 6 point type, next to the same image on an iPhone, and you’ll see why Apple’s math is not the end of the story. A child could see the difference.
But according to Apple it’s “done” now. Problem solved. They have the “math” to show it.
The flaw of Retina is the premise that we need to discern individual pixels. That’s not the only metric.
Also, I think Apple underestimates how close we hold the devices to our eyes, in certain situations (on the sofa, those in-between reading glasses and no reading glasses folks, etc.)