Picture Quality: Causes & Solutions

  • Thread starter Thread starter Tom McCurdy
  • Start date Start date
  • Tags Tags
    Picture
AI Thread Summary
The discussion centers on the analysis of a photograph taken with a 3.1 megapixel HP digital camera, speculated to have been captured with a long exposure setting. Participants suggest that the image's unique light patterns are likely due to camera movement during the exposure, as well as the motion of vehicles in a street scene. The presence of streaks and dotted lines in the image indicates both vehicle movement and camera shake, with the exposure time estimated to be under a second rather than minutes. The conversation highlights the complexity of interpreting motion in photography and the effects of both camera settings and external factors. Overall, the contributors provide insights into how the photograph's characteristics were achieved.
Tom McCurdy
Messages
1,017
Reaction score
1
What caused my picture to turn out the way it does. I really like it.
If you would like I could upload a full res version to my server
 

Attachments

  • lines.JPG
    lines.JPG
    28.6 KB · Views: 507
Physics news on Phys.org
Server location

Full res version
http://www.quantumninja.com/hw/random/IM000538.JPG
 
Not sure what the picture is of, but this smells like a low shutter speed.

What kind of camera do you have - manual/automatic/hybrid ? I suspect it's not a manual, or you wouldn't ask this question. Does your camera have a long exposure mode (usually for low light pictures taken off of a tripod) ? You may have inadvertently set it in this mode.
 
What is the picture of? The background kinda makes me think its of the night sky. A camera with a long exposure time could create a picture like that, and the lines would be the paths of moving light sources, but even still what moving light source would make those patterns?
 
A good portion of the "motion" is probably camera motion during a long exposure.
 
Yeah but sitll its pretty freaky, there are 3 main groups of light patterns so there would have to be some motion. The picture could have been a result of 3 different point of views all captured during the same exposure?
 
This is a street scene with cars.

The strongly foreshortened trapezoid along the bottom of the scene is the road. The red and yellow lights are receding cars, the white lights are approaching cars, the bluish blurs are streetlights.

Parallel or nearly parallel paths are caused by camera movement, probably over 10 seconds or thereabouts. The paths are not completely parallel because there are three degrees of rotation in the camera movement, affecting different parts of the picture differently.

The nonparallel streaks are caused by vehicle movement.

What's really cool is that you can see some lines that are not solid - they are dotted. Why? Those are turn signals!


It is surely taken with an automatic camera. Undoubtedly, the cameraman went to take a shot but did not realize that the camera's light meter told it to use a long exposure. The cameraman snapped the shutter, then continued about his business, unaware that the shutter was still open. Steady shot, followed by small motions, followed by larger and larger motions as he put the camera away.


The "three main groups of lights" are quite likely all part of one continuous line that passes in and out of the frame as the camera is moved.


BTW, with enough care and attention, you could probably recreate a good approximation of the scene, including car movement, camera movement and duration. I'd start with the streetlights. Since we know they don't move, all the blurring must be caused by camera motion. Once that component is normalized, you'd be able to reconstruct the vehicle motion.


Here is a
Street Scene with vehicle motion, but no camera motion.
 
Last edited:
DaveC426913 said:
This is a street scene with cars.

Good job!

But the total exposure time seems to be of the order of minutes.
 
Most of the light paths stay in a fairly regular arrangement to each other, even as they move across the focal plane. The lights only diverge from each other by about 10% (though it *looks* like it diverges a lot more). A line joining three arbitrary lights only grows from 820 pix to 920 pix over the course of the exposure- ~10%.

Look at this


Most of the movement of the camera (beyond to the shakiness) is vertical at first, then more horizontal. Additionally, the most promounced movement is near the end, while it is fairly steady near the beginning (see by the length of the flashing 'dots', LED lights, strobing somewhat faster than human perception.).

So, we have a cameraman holding the camera steady while taking the pic. But the shutter stays open. He thinks the pic is done and drops the camera to his lap, while turning it to the side (his right side).

So, we see streaks hold for a moment, then rise and disappear to the left.

The slight divergence (the image, as a whole, gets larger as the exposure progresses) is because the cameraman is in a vehicle that is moving. The shakiness too is from the vehicle.

This also means that the exposure can easily be quite short, a couple of seconds or less, not enough to show much movement of the vehicle. This is more in line with the speed that an automatic camera would try to do.
 
Last edited:
  • #10
DaveC426913 said:
This also means that the exposure can easily be quite short, a couple of seconds or less,...

Not exactly.
Count the number of transitions ( ~150 on-off ) along a "turn signals" path, and figure out the exposure time.
 
  • #11
I am convinced they are NOT turn signals.

1] Let's say blinkers flash at about 2 per second. That means those blinkers are on for more than a minute.
2] Not ONE of them is only on for a few blinks - which is how the large majority of people use them.
3] There are too many of them to be plausible.
4] They come in different colours - red, yellow, blue-green and lime green. They should all be yellow.
And finally
5] Each gap between adjacent blinks on that image would represent ~1/2 second. That means the camera, between each blink, is moving very slowly and VERY smoothly - without a single quick jerk, even a hand shake. Look at the scale of the motions vs. the proposed time frame (one blink per half second) Even the jerkiest segment has been represented by a dozen blinks. That means the fastest jerk of the camera took ~6 seconds.
And the jerks are in a complex sequence of directions.
And the camera would have had to behave this way for more than minute.
 
  • #12
Well it has been fun to see you guys talk about the picture let me share with you some information about how the picture was taken

The picture was taken with a 3.1 megapixel hp digital camera on the road in boston by yours truly. The exposure for the camera is a bit slow but no where near minutes. I would strongly guess the exposure time to be under a second.
 
  • #13
... :smile: ...
 
Back
Top