Not all at once

Hmmmm, I have a choice between a rant semi-continued from another post, or a kinda-long exposition that explains some curious traits. What to do, what to do?

strange lightning picture not mineI would like to give credit for this photo, but it’s one of the millions on the internet that were lifted from somewhere and make the rounds without attribution – oh, you internet! Anyway, it looks pretty bizarre and definitely puzzles a lot of people, but it’s actually not hard to determine how it came about, provided you know a simple trait about some cameras.

Despite the teeny tiny little lens and digital sensors on most smutphone cameras, they don’t operate in the same manner as DSLRs or film cameras, and don’t take the entire image at once. Instead, they have a ‘scanning’ method of recording the scene, very much like those old flatbed scanners, or a photocopy machine, or robots in cheesy science fiction depictions: the image is recorded a section at a time, generally running from one side of the scene to the other. It goes pretty fast, but if something is moving pretty fast (like a propeller) or happens very briefly, it gets captured in different portions of the action – this is worse if the light is low and the sensor is trying to get the most light from the scene. So what we actually see here is just a photo of the car, but during the exposure time, a very brief flash of lightning occurred right in the center of the frame – bright enough to illuminate the sky and, very slightly, the tree in midground, but short enough to only get captured during a portion of the recording scan. If you look very close, you can see that one arm of the lightning bolt to the right side persisted into the next block of the scan after the brightest light from the sky had faded. It’s simple, really.

Except that what I said earlier is a little bit of a misnomer, because even the better SLR cameras, digital and film, do this too, and have been for quite a while – anything with a focal-plane shutter does, which means anything with a shutter that sits over the film or sensor itself, rather than within the lens assembly (which is a fairly rare subset of camera types anymore, usually large-format cameras.) There is a physical limit to how fast a shutter can move aside to expose the sensor/film behind it, and this limit provides a certain set of restraints in photography – and some clever bits of engineering.

So let’s start with, there are actually two shutters (or to be more pedantic, two shutter curtains) within SLR cameras: one that moves out of the way from one side to the other to expose the media, and another that follows it in the same direction to close the opening. Nowadays with electromagnetic activation and very light materials, it takes roughly 1/300 second to completely cross the entire exposure area, but not too long ago it was about 1/100 second, and for medium-format cameras with much larger film frames and exposure areas, 1/40 to 1/50 second, give or take. And yet, for a long time, cameras have been able to get photos at much higher shutter speeds, 1/1,000 to 1/8,000 second. What sorcery is this?

Well, that’s the point of having those two shutter curtains. For a shutter speed of 1/4,000 second, the first curtain opens up and moves across the sensor (which it will take about 1/300 second to fully cross,) while the second starts to close behind it 1/4,000 of a second after the first starts. This means there’s a narrow gap between the curtains that travels across the media – again, like a flatbed scanner, but a hell of a lot faster and with less grindy noises. Any given point on the media is exposed to the light for 1/4,000 second, but not all of the media is exposed at the same time, and there are no sound barrier explosions from within your camera body.

And this is why, if you’re using the built-in flash of the camera, it always sets itself to a particular shutter speed and no higher: the flash can only go off when both curtains are fully open, or you get an effect like the lightning here. For external flashes, generally they are set to synchronize and go off only when the first curtain has fully opened, and on older cameras there was often a special mark on a particular shutter speed, say 1/90 second; any shutter speed below that was guaranteed to have the frame completely exposed by the flash, while above it would not.

However, there are also dedicated flash units that can expose at higher shutter speeds, often called some variation of ‘FP flash.’ To operate properly, they must communicate with the camera body precisely, because they won’t fire off one flash burst, but a series of them like a strobe light. They’re going off for each section of the media uncovered by the two shutter curtains as they travel across the frame, so the faster the shutter speed (meaning the smaller the gap between the two curtains,) the more bursts are necessary to light up the whole frame. Flash units need a lot of power to fire off that xenon tube within that provides the light, and this means charging a capacitor, so when a lot of bursts are needed, less of the charge can be used for each and the light strength is commensurately weaker and carries less distance.

By the way, this strobing happens so quickly (you know, about 1/300 second give or take) that we don’t see it blinking with our eyes, but just register it as one flash. Yet it syncs up perfectly with the shutter curtains so there’s no gaps or overlaps. Isn’t technology cool?

So now let’s look at another aspect of this high-speed shutter fudging.

Republic P-47 Thunderbolt showing propeller 'warpage'
Source: Wikimedia Commons

When things are moving fast, like propellers or race cars, their movement gets captured in different regions of the media as the shutter travels. So no, this isn’t an especially high wind, but the whirling blades and the sliding curtains coinciding in weird ways. The Republic P-47 has a propeller that rotates counter-clockwise, seen from this angle, so this means the shutter curtains were most likely traveling from bottom to top in this image – think about it and it’ll make sense. You can see, on the topmost blade for instance, how the shutter catches the root of the blade at one position, but as the curtains travel upwards the blade moves to the left and ‘bends’ in that direction. Meanwhile, the two blades on the left side are traveling against the shutter movement, getting compressed in size and spacing.

Except – this is backwards, because the lens always throws an inverted image onto the media, so the shutter was actually traveling from top to bottom of the camera; most camera manufacturers opt to have the curtains travel the shortest distance across the frame, for obvious reasons, so this means a top/bottom travel rather than left/right, though some older cameras do have shutters of this nature. With a little more research I could probably tell you the rough shutter speed that was used, given that engine’s idle RPM…

[Time out for a little warbird trivia. I get this “Mohican” impression from the shirtless crew chief there, though I think it’s just his hat hanging from a chin strap, but notice the difference in clothing between him and the pilot. Even in Burma, where this was taken, the temperatures at high altitudes are well below freezing and the cockpits were neither pressurized nor heated, so bundling up was necessary. Meanwhile, you can see the four .50 caliber machine guns in each wing, staggered so that the ammo belts could fit next to each other, and even the slots under the wing to expel the spent shell casings. The sleek P-51 gets a lot of attention, but the P-47 was an impressive aircraft that could absorb a tremendous amount of punishment and keep flying – check it out sometime.]

By the way, from time to time you might see a photo of a race car, often from when someone can get right down to the fences, that is skewed diagonally. Again, trying to use a high shutter speed that doesn’t actually freeze the car in place, so it’s moving horizontally as the shutter curtains travel vertically. This is why panning the camera works better.

ice cube and water droplets in midairAnd one last trick, another way to overcome the limitations of a focal-plane shutter. In dark conditions, you can actually lock the shutter open, and use a flash as the sole light source, so the brief duration of the flash is the actual exposure time of the image (this usually ranges from about 1/1,000 to 1/10,000 second.) This is a method pioneered by the guy that developed the xenon flash tube, the heart of nearly all standard flash and strobe units, in the first place: Harold “Doc” Edgerton. He’s the guy that did all of those classic photos of the milk drop and the bullet passing through the apple. That kind of jazz takes some engineering skills with microphone triggers and careful calculations for the latency of the electronics, every step of the way, but in most cases you can get by, like here, with some careful timing, a bit of luck, and a lot of tries.

I mentioned a different type of shutter earlier, and that’s the iris shutter – it sits within the lens assembly and often doubles in function as the lens aperture, opening to a preset size to expose the film. Because of its shape and position within the lens, there are no focal plane or shutter curtain effects, and it can flash-sync at any shutter speed, but generally the shutter speeds are limited (physically/mechanically) to a maximum of 1/500 second anyway, and often none too accurate at that. All this means is that each lens has to have its own shutter assembly, increasing expense and complication, and this practice is primarily limited to large-format cameras. I’ll go into the pros and cons of those someday…

UPDATE 02/16/18: So, I decided to attempt to calculate the shutter speed of that plane photo above, and began playing around with it after determining the idle speed of the P-47 is 900 RPM. After a lot of fiddling around, I realized I wasn’t mathematically-inclined enough to figure it out. I mentioned this, however, to Jim Kramer while I was speaking with him on the phone, and he provided the answer while we were in conversation: allowing for some slop given inexact measurements from blurry propellers, the camera had a shutter speed of ~1/360 second. Knowing that virtually all cameras at the time would have had only 1/250 and 1/500 second settings, and that the mechanical shutters typically fell a little on the slow end, I figure the photographer had likely set for 1/500 second. I had been trying to determine the distance the blades had covered while Jim went with angle instead, but Jim’s an engineer. He probably just asked one of the passengers on his train…