Post date: Jul 16, 2012 1:47:33 PM

The correct answer is more interesting! It's such a shame that the first answer was wrong and nobody followed it up.

The effect blamed is oversaturation, known as CCD "bloom". You'll know it well if you've seen much TV footage from the 70s with bright lights pointed at the camera. Here's what CCD bloom looks like (each image links to the source page if you'd like to read more about it):

CCD Bloom. A real thing, but the temple photo effect is not this.

The short explanation is that CCD sensors pass an electric charge along each row of pixels when reading them out, but if the light along a few pixels is extreme compared to the rest of the image, those pixels oversaturate and too much electric charge is captured, and it then spills / smears out affecting the whole row. Rows with too much light are oversaturated (usually in both directions, but at least in one direction from the source of the saturation), and the effect is blurred out over a few rows.

So why can't the Mayan Temple photo be a CCD effect?

• Well it doesn't really look like CCD bloom. If you're familiar with CCD effects, that's immediately obvious. The "beam" is not oversaturated, which the most obvious problem if you're claiming oversaturation.

• The bright band has sharp edges, not blurred sideways like bloom.

• Furthermore, the brightest bit of the strip is not at the end of the banding, so the bloom doesn't seem to have a source.

More to the point though:

• iPhones don't use CCDs!

Very few modern cameras use CCDs any more (though that's a pretty recent trend). Small-format cameras like cameraphones haven't used CCDs in a while. They all use CMOS sensors - a different technololgy (with a different set of issues, one of which caused the awesomeness in the photo).

So it's not a CCD error - what the heck is it? A friend at work figured it out (Thank you Irrgy!).

The correct answer is something way cooler than bloom - and it's actually a bigger cooler coincidence that the beam matched up with the tip of the temple. What it is, is a

rolling shutter artefact!

Rolling shutter artefacts are awesome. In contrast to CCDs, CMOS sensors don't pass charge around, but they do only capture one row at a time, so there's a delay between capturing one row and capturing the next. This is known as a "rolling shutter" because instead of capturing the entire image at once, they capture rows in order, and if anything in the image is flickering or moving fast enough, it gets skewed. Take a look at this awesomeness!

(click for source)

The propeller is spinning, and the CMOS sensor in the iPhone captures one vertical column at a time. At the left and right of the propeller, the blade spins past the shutter as it rolls left to right, leaving a streak as it goes past. The least-bent blade appears sitting on the shaft as the shutter sweeps past it.

Confusing? I took this video of a slowly spinning propeller, and made a mock-up of how a fast propeller would be captured by a CMOS sensor with a rolling shutter:-

(To get the relative speeds right, I'm actually capturing three rows at a time which is why the edges are a little jagged - they're actually 3-pixel strips.)

It's not just still photos that are affected though! If you take a normal video (not a high-speed video) of a spinning propeller, then every frame gets affected. If instead of moving something fast, you shake the camera very fast, everything gets distorted. What fun! Search around for "Jellycam".

If you have a Canon DSLR, you can use Magic Lantern to artificially slow the effect right down take interesting "slit-scan" photographs!

Well as you can see this is all pretty cool. But how does all of this relate to the Mayan photo?

Let's take another look:

It shows a bunch of normal sky to the left and right, the ground and temple are normal everywhere, only one patch of sky is lit up above the temple.

The lightning is the key.

Lightning is actually one or more short flashes and a glow - take a look at the end of this slow motion video. You can see the afterglow in the temple photo to the left of the bright area, which explains what happened (and tells us that the image was captured right-to-left). What happened is that the lightning lit up the sky while the camera was capturing that part of the photo. I used the same script as for the propeller, to make this video:

And you can clearly see how different parts of the sky are captured at different brightnesses.

If you take a video with a simpler example lightning strike (fewer leaders, only really one main flash), you see something much closer to that temple photo. To labour the point I've made it a right-to-left scan too:

Besides looking more correct, it also explains why the temple and the ground look fine; the lightning is far away and behind the temple, so it doesn't change the appearance of the temple or the ground, just the sky behind the temple. In the example above, the lightning is closer and much brighter, so some of the buildings on the pier are slightly affected, but not much so the difference between bloom and rolling-shutter brightness is still clear.

With a little Googling, I found a video with lots of frames affected in the same way, and a nice example photo:

The photo shows the banding effect more clearly, though in this photo the lightning coincided with the area of the frame capturing the lightning. This is less coincidence and more pragmatic - the lightning doesn't affect that much of the sky, you can barely see the line at the right of the image. If there were other strikes in the sky while the ground and road were being captured, it just wouldn't really be visible.

Coooool, huh? The lightning strike was perfectly timed with the capture of the temple in a rolling-shutter CMOS sensor.

Now what are the odds of that?

Maybe it's a sign of alien interference with the camera...