Sense & Rhythm

Notes on design, film, and tech by Phil Bolles

[senseandrhythm]

Touch Bar Refresh Rate: A Note on Strobing

Ever since I got the MacBook Pro with Touch Bar, something’s bothered me about the way its display works: it strobes on and off rapidly.

With a refresh rate that’s quite high, about 240 Hz, it may sound like it's smooth, but it's really not; most of the time, it’s not displaying anything.

Above is a slow-shutter photo at 1/20th of a second, with a moving camera, and the Touch Bar cycles about 11 or 12 times during that period and doesn’t show a blur -- compare the lack of motion blur to the smooth blur on the keyboard and screen. 

Most people probably won’t notice this strobing, but it feels like a bit of a defect to me.

[senseandrhythm]

Back to Basics

From time to time, I've found it hugely beneficial to re-tread a territory that I thought I knew; I almost always find something I didn't, in fact, know.

Even though I've been coding CSS for many years, I recently went through James Williamson's Learning CSS series and was impressed by its detail. It reinforced my knowledge of styling, but it also reframed it in some ways.

For instance, the point values of specificity: I didn't know that this was easily quantifiable in a four-digit value. Targeting an ID gives a rule a value of roughly 100 (or 0,1,0,0... since these aren’t really base-ten values) vs. targeting just an element, which is only worth 0,0,0,1.

As soon as I saw that, it was like a new connection formed in my head, tying together hierarchies, rules, and counting methods. I started to wonder what a system is called when its base is infinite, but there’s still an order of magnitude, like hexadecimal but instead of 16 values... as many as you want.

It was one of those days where I actually needed to go back to the W3C specification and confirm it, check out its context (and try to hopefully remember it). I had known about specificity, but I didn’t understand how easy to quantify it was.

It's great when you realize that you've found a missing piece to a puzzle, and I love being able to have a more fully formed understanding of a subject.

So, here's some encouragement to all of you out there who want to learn something better -- go back to the basics, think about it differently, find the missing pieces, and reframe your mental models.

(Image based on the W3C CSS 2.1 Spec)

[senseandrhythm]

Large Format Follow-up: An Unexpected Focus Issue

In May, my friends and I did a second shoot with the Eastman View No. 2 5x7 large-format camera. 

Above is an image of my friend Kyle---and though it’s basically acceptable focus, it’s slightly off. Looking more closely at it, I noticed that the focal plane seems slanted.

Never having seen this before with other cameras, I had to think about why this might be the case. Kyle’s shoes show that the focus is slightly too far back and not parallel to the image plane.

Connecting the points that are acceptable focus, we can see that the focal plane is angled---it’s something like this:

Why would the focal plane not be parallel to the image surface? In a word: tilt/shift. My lens isn't guaranteed to align perfectly with the backing; manual adjustments are necessary.

Guess I'll have to double-check their being parallel in future shoots.

Bonus image: a photo of yours truly, taken by Kyle.

[senseandrhythm]

First test shoot with a 100-year-old, large-format camera

It’s a shame when an old camera in working condition is merely decorative. Shelves of photography stores are lined with cameras that are, for all intents and purposes, usable. For the past few years, I've made a goal of using any antique camera that comes my way.

Because large-format photography has been an interest of mine recently---due to its required intentionality, its high image quality, and its strong connection to history---I decided to research its process and find a camera to learn with.

I browsed through used listings online, watched tutorials, and read articles. Slowly, I began to gain some familiarity with the details that make the large-format process different from using 35mm or digital SLRs.

And now, I actually have a camera of my own to shoot with---a few weeks ago, I received an Eastman View No. 2 5x7 field camera (circa 1914) as a Christmas gift.

A brass plumb-line and new fabric bellows

It's been refurbished over the years, with a new ground glass, spring mechanism, and newly created fabric bellows. But the cherrywood/brass construction of the body is original and I'm fairly sure that the lens and shutter are, too. Looking it over the first time, I got the sense that it was fully functional.

Detail of brass work and plate below the lens board, on the front standard

Detail of brass work on the rear standard

Unfolding and setting up the camera the first time was fairly easy, however I couldn't figure out a couple of its parts.

Although I've studied photography for about twenty years, I know that there are gaps in my knowledge, especially when it comes to large camera systems with bellows and standards.

And over the past month or so, the more that I read, the more it was clear how much I didn't know.

Diving back into books, articles, and even YouTube slowly helped to bring me up to speed.

ASK QUESTIONS FIRST…

I was now able to answer many of my initial questions such as, "How does the ground glass work?" (Adjust focus on the glass and compose your image before you load the film.)

And I found answers to my other questions, such as:

“How do I load film into the camera?” (You need special film holders—each one only carries two sheets of film.)

“Once I load film, can I still use the ground glass?” (No. The film blocks the ground glass, so you’d better get the composition and focus right before adding film.)

The Ilex shutter’s fastest exposure is very slow by today's standards: only 1/150th of a second

A marking reading “8 ½” hints that the lens is old; its focal length is measured in inches

And eventually, I answered more questions like: “How does the shutter work?” “Are there any light leaks?” “Once I fire the shutter, where do I store the film until I take it to the lab?” “How much will it cost for the film, processing, and scanning all together?”

Needless to say, I had several issues to figure out.

And I needed supplies. For essentials, I picked up sheet film (Ilford 400ASA black and white), several film holders, and a changing bag---the combination of knowledge and materials finally made me ready to shoot.

SHOOT LATER

And so, with assistance from a couple of my friends, I started shooting with the Eastman.

Since my Ilford black and white film is rated 400 ASA, and since the lens on my Eastman opens to f4.5, I knew I could figure out which shutter speed I needed in order to get a good exposure.

I used my digital SLR to take test shots and light readings---it told me I needed to shoot at 1/200th of a second. Since the Eastman's Ilex shutter only goes as fast as 1/150th, I used that speed (which is close enough) and exposed some negatives.

One of my first images created with the Eastman 5x7

Getting the negatives back from the lab was a huge relief. The camera worked, and our exposure tests were accurate. I loved seeing the detail, the depth of field, the wide dynamic range... all right there on film.

It's great when a method is actually repeatable and gives you predictable results (even after 100 years)---indeed, it’s one of the amazing things about science.

I’ll be creating and posting more photos in March and April; I look forward to shooting with the Eastman View in other settings and continuing to learn along the way.

[senseandrhythm]

Scaling in Siri's Suggestions

From the iPhone's home screen on iOS 9.1, swiping right brings you to the search screen, which also shows "Siri Suggestions" (also, swiping down from the home screen shows a similar interface). Essentially, it's a list of people you may want to contact and apps you may be interested in using.

Peculiarly, this “suggestion” screen shows versions of app icons that aren't quite the same as on the home screen. It's almost as if they've been scaled, except that they're the same size overall, just a different resolution.

Fine detail is lost, edges are fuzzier.

[caption: On the left, the home screen version of the Settings icon; Siri's fuzzy version of the same icon, right]

And it's unclear why this happens at all.

[caption: the two versions overlayed]

My hypothesis is that iOS loads a low-resolution version of the icon for speed/performance reasons, and scales it up to be the correct visual size.

It will be interesting to see how this behavior changes in forthcoming versions of iOS 9.

[senseandrhythm]

A Macro Look at Anti-Aliasing and Sub-Pixels

Drawing smooth diagonals and curves on a screen should be easy. But without anti-aliasing, it wouldn’t be possible. Smoothing out lines, curves, edges—or allowing the user intentionally not to do so—is a mainstay for graphics creation on the screen.

It’s easiest to see how anti-aliasing works by looking at just one channel.

As a review, most monitors use additive color systems of three channels: red, green, and blue (RGB). Added together, these three channels create white. When two of these three channels are combined, they create the secondary colors cyan, magenta, and yellow (CMY).

In order to look at single channels, I’ve drawn out shapes in Photoshop colored with 100% red, 100% green, and 100% blue.

CAPTION: Overlapping shapes in red, green, and blue (no anti-aliasing) on left. On right, these shapes are redrawn with anti-aliasing turned on.

Pictured are three overlapping colored triangles, drawn in Photoshop on a Retina Macbook Pro 13.3” (2015) with blend modes set to “Lighten.”

Macro photography helps us observe the difference between the two. Above are photos of the Macbook Pro’s screen—clearly identifiable are the sub-pixels (in red, green, or blue) that together make up whole pixels. Each full pixel contains all three sub-pixels, and each pixel is less than 1/200th of an inch wide on this particular screen.

Where there is a pure color that is displayed using just one channel, e.g. in the shape that’s 100% red, there is only one sub-pixel lit up. The other two sub-pixels that would be on the right of it aren’t lit up at all; those green and blue ones are at 0%.

Looking at the diagonal edges of the red shape drawn without anti-aliasing, there’s a jagged, sharp cutoff on the edges.

CAPTION: For these sub-pixels, “A” is at 100% while “B” is at 0%.

This is because the sub-pixels are either “on” or “off” when shapes aren’t anti-aliased, and because the grid of a computer monitor isn’t well suited to represent diagonal lines or curves.

CAPTION: Sub-pixel “C” = 100%; “D” ~= 72%; “E” ~= 41%; “F” ~= 10%; “G” = 0%.

In the anti-aliased version, the sub-pixels taper off in brightness, from 100% down to 0% over multiple pixels---this tapering creates a softening effect on the diagonal line.

Single channels now hopefully make sense. As an example of multiple channels, let’s look at white text against black.

Text shown here is set in Avenir. On the left, there’s no anti-aliasing used, while the image on the right shows the “Sharp” anti-aliasing setting within Photoshop’s character palette.

Up close, the difference between the two is more noticeable. Just like the red sub-pixels we previously looked at, the sub-pixels for the letters that don’t have anti-aliasing are either “on” or “off.”

Just the letter “e” by itself shows a clear difference. For readability, it’s often preferable to use anti-aliasing when text is involved.

OS X and Windows both have default anti-aliasing, and many applications (e.g. Photoshop) allow the user to choose whether to anti-alias and how much to do so (and when to disable it, which can be very helpful when one needs pixel-perfect accuracy).

Overall, anti-aliasing is a key way to facilitate making smooth shapes; understanding what it is and how it can be used is a secret weapon for any digital artist’s arsenal.

[senseandrhythm]

4K Problems

It’s the next jump in resolution after HD, but 4K (or UHD, which is about 3.8K) presents challenges when it comes to content and technology.

Besides the question of whether a person can tell the difference, i.e. a real, perceptual improvement in resolution---see Phil Plait's article on arcminutes and 4K TV---there are two main problems with the distribution of 4K films: 1) Acquisition, and 2) Visual Effects.

Let's deal with the acquisition question first.

Movies have to be shot in an ultra-high-definition format in order to be 4K. There aren't enough 4K films in existence to put together a library of any consequence.

When movies originate on film, each frame of film is scanned at a certain resolution, like 2K, 4K, or even 8K---whichever is deemed adequate by the production company or studio, and 35mm film is an ultra-high-def format.

But most movies aren't shot on film anymore, so they can't arbitrarily be scanned. Instead, their resolutions are dependent on the digital systems used for acquisition; these camera systems usually have a resolution of about 2K or 3K.

To screen in 4K, digitally-captured movies are often scaled up to a 4,096-pixel master or simply mastered in 2K (2,048 pixels across). A few films, such as Girl with the Dragon Tattoo, were shot and mastered at higher resolutions like 4,352 × 2,176, but they extracted from that master a final frame size of 3,600 x 1,500 pixels, still less than 4K.

But even movies shot on film haven't traditionally gotten the 4K treatment. Lord of the Rings: Fellowship of the Ring was shot on 35mm film (Super 35), scanned for a digital intermediate, and mastered in 2K---but since each frame is a 2.40:1 widescreen extraction, its final, mastered resolution was 2,048 x 854, which is only 2K and vertically much fewer pixels than even full-HD.

Twenty years ago, before digital cinema cameras were commonplace, there was another form of digital origination starting to emerge: rendering the entire film within the computer.

Toy Story was the first all-digital 3D animated movie, and it was rendered at less than 2K---only 1,536 x 960 pixels.

Let’s turn to the second issue: visual effects. Even for movies that have 4K masters, almost none of the effects are in 4K. Current estimates suggest that visual effects are rendered in full 4K for less than 5% of films; the vast majority of films render VFX in 2K.

Multiple reasons are behind this fact: time, cost, bandwidth, even the textures in the VFX CG models themselves which would need to be made at higher resolutions to render correctly in 4K.

One aside on color: UHD specs recommend a 10-bit color gamut (30-bit color total, compared to traditional 24-bit color distribution formats). And wider color gamuts mean larger amounts of data. A UHD frame is four times the data in pixels, but it's also 25% more data in terms of color-depth.

An example: one uncompressed frame of 10-bit UHD is 31.1 MB (3840 x 2160 x 3.75 Bytes), compared with 6.2 MB per 8-bit HD frame (1920 x 1080 x 3 Bytes).

_____________

5x the data 31.1 MB/frame vs. 6.2 MB/frame _____________

So, UHD is about 5 times more data than HD---even if they're compressed using similar codecs. And if the UHD compression schemes are able to be twice as efficient as those used in HD, there will still be a challenge in distributing 2.5 times the data; bandwidth alone will be an issue, as it has already started to be for subscribers of some streaming UHD services like Amazon and YouTube.

After all of these issues are taken into account, the upshot is that 20 years' worth of cinema (from the '90s to the '10s) are in 2K or less than 2K, and they always will be so. Their content will be scaled up to fit the screen, and even brand-new movies will incorporate 2K effects inside of 4K-acquired content.

A final note: From the way 4K TVs are advertised, you'd think there would be plenty of content for them. We know that this isn’t true in terms of films. Sometimes, 4K TVs are targeted to gamers, as in this recent ad from Best Buy:

Best Buy would have you believe that the XBox One supports gameplay at 4K; it doesn't.

Graphics from the games are scaled up from HD to 4K—and in that case, what's the point of a 4K display? This situation is like buying a new car: you've got a huge, V8 engine, but your fuel line can only supply enough gasoline for four cylinders, leaving the remaining four motionless.

Unfortunately, for both films and games, this kind of fuel starvation is what we're seeing in the 4K world.

Further reading:

IBC: 4K and better pixels

Why Buying A 4K TV Right Now Is A Waste Of Money

4K on the Cheap: Ultra HD TV Is Almost Ready for Mainstream

4K TVs Are Forked. You Should Wait Before You Buy One

[senseandrhythm]

Dashboard Cursors

Bring up a Mac Dashboard containing more than one sticky note and you’ll find that each one prompts input with a flashing cursor, regardless of which one is selected. Apple’s decision is curious in this regard---I would recommend prompting input only for the most recently selected sticky window, not all of them.

Even after the user types into the window, all three cursors continue to flash. This seems like a less than optimal user experience, though I’ll be experimenting with more use cases to see if I can figure out why they decided to build it in this way.

[senseandrhythm]

In between clicks, an (almost) analog feel

My quick observations on the near-analog nature of the Apple Force Touch trackpad:

Transcript:

A lot people have been talking about this new pressure-sensitive trackpad on the MacBook Pro and the MacBook. It turns out, my new MacBook Pro (13.3 inch) comes with it; so, I figured I would give it a shot, test it out.

So, people have written about how you can actually position the cursor over a file, you know, click once, but then when you click a second time, it brings up the QuickView. Now, you can do these really in quick succession---like, you do "first click," "second click"---you feel two clicks. It's almost like you're clicking *through* the first click. I'm going to get rid of that [the QuickView].

You're clicking the first time and then through the first click into the second click. But what I haven't seen people talk about a lot recently is that, for instance, if it's not selected, hover over something, you have that first click... and then in between the first and second click you get a sense of the pressure-sensitivity.

Now, you can be seeing this thumbnail kind of increase, decrease in size---this is actually in between the first level and second levels of the click. Gives you a really good sense of how much pressure sensitivity is actually in this trackpad.

It almost feels kind of like an analog... it's like the first time I got the analog joystick for the N64. It has so much, so many little levels; it almost feels very fluid...

Here, I'm clicking a little bit lighter and then increasing it. And then at a certain threshold, it'll pop into the second click. There---so, then it just clicked into the second level. It almost feels like a deeper click than the first one.

And then, of course, the video plays as normal.

So, what I also should show you is if you double-click this video to open it in Quicktime, you can start playing the video, and then you can hold just a little, a faint click is 2x. If you click one more level it's 5x, one more level it's 10x. And that actually keeps going all the way through... from ten to thirty to sixty. And I think it ends at sixty, but I'll show you what that's like.

It's actually... what I'm feeling here is a slight click every time I get into a deeper level. So, five, another click for ten, another click for thirty, sixty, etc. Click down once, and it feels like you're clicking a second time, deeper, and getting further and further into the trackpad.

Of course, it's mostly an illusion because it's really using electromagnets to actually fool your brain into thinking it's kind of getting to a deeper level.

It's very cool, and it's definitely the future, enabling multiple levels of trackpad sensitivity---multiple levels of click---and actually giving you some force feedback every time you enter a deeper level.