Why is my world (as seen through my phone) distorted?
July 31, 2010 5:33 PM Subscribe
Problems with parallax... using a Google phone to video stuff in motion produces a result that I can't explain...
I took a video while on a train journey in Japan. As I pass by vertical objects, they start vertical on the leading edge of the picture(towards the direction of travel), but by the time they get to the trailing edge of the picture, they appear to tilt about 15 degrees or so towards the direction of travel. However, this is ONLY TRUE for items near me. Items far away are unaffected.
A still from the video is here. The nearby telegraph pole is tilted, but the distant one isn't. Thoughts? Explanation?
I took a video while on a train journey in Japan. As I pass by vertical objects, they start vertical on the leading edge of the picture(towards the direction of travel), but by the time they get to the trailing edge of the picture, they appear to tilt about 15 degrees or so towards the direction of travel. However, this is ONLY TRUE for items near me. Items far away are unaffected.
A still from the video is here. The nearby telegraph pole is tilted, but the distant one isn't. Thoughts? Explanation?
(it's not a "typical wide angle effect" as your flickr commenter indicates)
posted by fake at 5:51 PM on July 31, 2010
posted by fake at 5:51 PM on July 31, 2010
Isn't the rolling shutter supposed to produce a "rolled" bend in things? I don't see that here.
posted by circular at 6:22 PM on July 31, 2010
posted by circular at 6:22 PM on July 31, 2010
Watch this demonstration clip -- the straight vertical lines become oblique; no rolling involved.
posted by Rhomboid at 6:49 PM on July 31, 2010
posted by Rhomboid at 6:49 PM on July 31, 2010
Isn't the rolling shutter supposed to produce a "rolled" bend in things? I don't see that here.
This isn't my experience. If you're looking at this "rolled" helicopter, think about how the blade moves as the CCD/shutter captures the image from top to bottom: the tip of the blade (which has the furthest to travel) is moving the fastest, and the speed decreases as you move down the blade toward the axis of rotation--because the entire blade is (of course) revolving at the same rate. If the shutter is scanning down the image at a steady rate, the distortion will decrease as the speed of the moving blade decreases, producing a "bend" rather than the "skew" seen when objects moving at a uniform velocity are photographed (e.g. this car)
I think.
I always assumed that the "rolling shutter" was so named because some early focal-plane shutters were made of opaque fabric which was literally rolled top-to-bottom in front of the film or plate, exposing it.
posted by pullayup at 6:55 PM on July 31, 2010
This isn't my experience. If you're looking at this "rolled" helicopter, think about how the blade moves as the CCD/shutter captures the image from top to bottom: the tip of the blade (which has the furthest to travel) is moving the fastest, and the speed decreases as you move down the blade toward the axis of rotation--because the entire blade is (of course) revolving at the same rate. If the shutter is scanning down the image at a steady rate, the distortion will decrease as the speed of the moving blade decreases, producing a "bend" rather than the "skew" seen when objects moving at a uniform velocity are photographed (e.g. this car)
I think.
I always assumed that the "rolling shutter" was so named because some early focal-plane shutters were made of opaque fabric which was literally rolled top-to-bottom in front of the film or plate, exposing it.
posted by pullayup at 6:55 PM on July 31, 2010
Incidentally, I believe that this is often why fast things in cartoons appear to lean forward.
posted by pullayup at 7:07 PM on July 31, 2010
posted by pullayup at 7:07 PM on July 31, 2010
Best answer: Isn't the rolling shutter supposed to produce a "rolled" bend in things?
No. The name refers to the way the sensor is read out, "rolling" from top to bottom, not to any image artifact. The primary image artifact in the linked image is skew, and it is pretty well known.
posted by fake at 7:21 PM on July 31, 2010
No. The name refers to the way the sensor is read out, "rolling" from top to bottom, not to any image artifact. The primary image artifact in the linked image is skew, and it is pretty well known.
posted by fake at 7:21 PM on July 31, 2010
Response by poster: @fake - thanks! I think you have it. I first noticed the effect with a train that went past. Since it was nearer, it was even more pronounced, but I have only just managed to find the frame and grab it.
posted by blue_wardrobe at 1:27 AM on August 1, 2010
posted by blue_wardrobe at 1:27 AM on August 1, 2010
Great stuff. You should be able to get the same effect by moving your camera rapidly in a line -- just set it on a countertop and slide it quickly. You'll see the world distort accordingly. Cheers on your discovery -- cameras have lots of interesting idiosyncrasies. I keep thinking that one of these days, someone will figure out how to exploit rolling shutter artifacts to do something interesting...
posted by fake at 6:35 AM on August 1, 2010
posted by fake at 6:35 AM on August 1, 2010
I keep thinking that one of these days, someone will figure out how to exploit rolling shutter artifacts to do something interesting...
Like this?
posted by Mwongozi at 7:32 AM on August 1, 2010 [1 favorite]
Like this?
posted by Mwongozi at 7:32 AM on August 1, 2010 [1 favorite]
This thread is closed to new comments.
Objects that are closer to you move more quickly across the sensor plane, and therefore, they distort more. Objects which are far away move more slowly, and, if they move slowly enough (which is the duration of one frame of capture), appear not to distort at all.
This is a major problem in digital sensing today. Most cameras exhibit some form of rolling shutter artifacts. But before anyone goes ragging on about how film was better, it wasn't, necessarily.
posted by fake at 5:50 PM on July 31, 2010