"Photon gating" makes for interesting cameraphone pictures

A friend sent me this image that he says isn't photoshopped. The distortion of the airplane props are due to the way a cameraphone's low-end CMOS sensor records an image. Wired explains:

The iPhone has no physical shutter and instead uses photon gating on its CMOS sensor. Some parts of the image are recorded before others, much like with a scanner. The iPhone's CMOS scanner seems to be a lot slower than, say, the CMOS sensor on your Canon point and shoot camera. Therefore, as the camera is recording the image, any changes over that small but significant amount of time are recorded.

[#1 posted by mr.skeleton, January 13, 2009 10:53 AM]

It's almost... impressive (?) how slow the iPhone camera is. I was turning the phone from landscape to vertical because I thought it had already taken a picture and got this sort of neat effect:

http://www.flickr.com/photos/mr_skeleton/3129571418/in/set-72157611465247733/

[#2 posted by Alan, January 13, 2009 11:17 AM]

Another use for the iPhone!

[#3 posted by Sxe, January 13, 2009 1:57 PM]

Older digital cameras have the same effects.

There have been a few artists who've achieved these effects by moving pictures/objects around on a flatbed scanner.

Example:

http://hurraylien.com/images/jordscrew04.jpg

[#4 posted by imh, January 13, 2009 2:02 PM]

You don't even need to photograph anything fast, just set the cameraphone to take a picture and then throw it into the air. The inevitable spinning sometimes makes a great shot...

http://www.flickr.com/photos/mongolblues/265489553/in/set-72157606618190079/

[#5 posted by JoeKickass, January 13, 2009 2:02 PM]

This also works with non "i" phones.

[#6 posted by Anonymous, January 13, 2009 4:26 PM]

Anyone willing to do the math about what works best with this shutter speed? I think the entire scan usually takes at least a 30th of a second or longer.

I have had intresting shots shooting from a moving vehicle.

What you would need is something moving past at set rates, for either direction of the image. Something strobing at a known rate.

[#7 posted by airshowfan, January 13, 2009 7:45 PM]

This also happens on old cameras with curtain-style shutters, which also "scan" across the frame, kinda. Many aircraft pictures from World War 2 show this effect - although not quite this visibly. I'll go dig some up and post some links here.

[#8 posted by airshowfan, January 13, 2009 8:00 PM]

I could only find one example taken with an old film camera, where the movement of the shutter across the frame as the prop rotates causes the blades to look bent...

http://en.wikipedia.org/wiki/File:B29.maxwell.750pix.jpg

... but there are tons more like this around the interwebs. They can be found by Googling "B-29", "Hawker Hurricane", etc.

And here's a thread on Airliners.net about this:

http://www.airliners.net/aviation-forums/general_aviation/read.main/4278311/

Here is the effect on a video of a prop spinning up:

http://ca.youtube.com/watch?v=vlZvBIgTE68

[#9 posted by airshowfan, January 13, 2009 8:07 PM]

Ok, just one bunch of examples:

http://www.fencecheck.com/forums/index.php/topic,6032.0.html

[#10 posted by Anonymous, January 13, 2009 9:28 PM]

This works on non "i" phones too. as long as you have a slow camera. This was taken on my old Nokia 6680 from inside a moving veicle: http://www.flickr.com/photos/silvr/19357487/ the distortion caused everything to slant something crazy.

There's an old technique related to this which is known as slit-scan photography. Applied to video, this gives rather creepy results too. (search YouTube)

[#11 posted by georgelazenby, January 14, 2009 12:02 AM]

math geeks will note that it looks like a tan function.

[#12 posted by kenturamon, January 14, 2009 4:52 AM]

I have actually been able to use this effect to time a flash manually to be picked up the camera, a bit like this: http://flickr.com/photos/kenturamon/2531198987/

That time I only caught "half" the flash, but it shows the effect nicely.

[#13 posted by Cyclelicious, January 14, 2009 9:47 AM]

Wonderful symmetry in that photo!

iPhone users call these wavy photos "Jello Shots."

[#14 posted by Anonymous, January 14, 2009 10:10 AM]

When I was a rocket scientist, we called this sort of thing "aliasing" when it occurred during digital data acquisition. We worked rather hard to make sure our data sampling intervals didn't alias cyclic phenomena. For example, if you are measuring exhaust gas pressure from a statically mounted rocket motor at 100 samples per second, and the pressure fluctuates 400 times per second, you will see at flat line instead of a 400 cps wave... I'm not going to look up Nyquist and all that right now, I think y'all will get the idea.

Generally, you should sample more than twice the frequency of anything you might be interested in. If you're doing pure science (or quality control) sample at the highest economically feasible rate and consider doing pure analog recording and analysis as well.

I like to use prime numbers when I'm scheduling duty cycles... why trigger events every 60 seconds when you can do 61? Why hit your server every 15 seconds when you can do 17? Using primes spreads traffic better.

--Charlie

[#15 posted by Anonymous, January 14, 2009 7:15 PM]

"Generally, you should sample more than twice the frequency of anything you might be interested in. If you're doing pure science (or quality control) sample at the highest economically feasible rate and consider doing pure analog recording and analysis as well.

I like to use prime numbers when I'm scheduling duty cycles... why trigger events every 60 seconds when you can do 61? Why hit your server every 15 seconds when you can do 17? Using primes spreads traffic better."

Ooh Nyquist Sampling frequency. That brings back memories... :|

Very interesting idea about the prime-numbered duty cycles. I'm going to remember that the next time I poll anything.