In this video, I see some sort of artifact (don’t know if that’s the right word?) apparently caused by the propellor blades. It seems to change based on RPM. I didn’t shoot this clip, but I’ve seen this before in videos thru the propellor blades of a plane, and I don’t know what causes it. Can any of our Doper experts shed some light on this?
FTR: I’ve got a new student at the airport (hard to pick out the field 'till near the end of the video). I think it’s a cool place to teach. Kinda like going back in time.
I figured something like that, maybe similar to the way spokes seem to move backwards in celluloid film sometimes. But I’m puzzled the strange curves of the propellor blades. Is that caused by the way the CCD (CMOS?) chip captures the scene? Does it capture the scene somewhat in the same manner that a TV tube constructs it? (line by line?).
I’ve actually seen this in a still photo taken by a digital camera. I think it’s unique to digital cameras.
Specifically, the pixels don’t all grab an image at once like an idealized physical shutter. They are each exposed in turn, so by the time the last few get a peek, they’re looking at the same propeller blade that the first ones saw, but in a different place. Here’s a good page on the effect: http://dvxuser.com/jason/CMOS-CCD/
The key here is that there is no shutter, hence no blurring. Shutter blur occurs when the object moves during the period of time that the shutter is open. For digital cameras, the values for a set of pixels are read instantaneously, not over a period of time.
That’s certainly not true of digital still cameras. I’ve taken many pictures with digital cameras that exhibit motion blur. Digital cameras still have a shutter, and the sensor is exposed for as long as the shutter is open.
I’m sure I’ve seen motion blur in digital movie cameras too.
Pixels are read over a period of time. The sensors allow light to build up change at the pixel for some amount of time and then read the amount of charge.
No measurement is instantaneous. We think of a lot of measurements as occurring at a specif point in time to make the math easier. This does not result in much error if the measurement happening quickly enough that the measured thing does not change much in the time span of the measurement.
I’m not an expert at this, but I think that the type of image sensor (among other factors) can affect blurring.
In the image I posted, I believe the less expensive CMOS sensor was exhibiting the ‘rolling shutter’ problem. My (limited) understanding is that this bizarre result occurs under conditions what would cause ‘blur’ in CCD sensors or film. So, perhaps digital cameras that produced blurry images are using non-CMOS sensors or the different ‘global shutter’ technique for capturing the image.
This happens to be my direct area of expertise!
CMOS sensors typically allow for electronic rolling shutter mechanism, where the actual readout of a pixel causes it to be shuttered. ERS often allows very fast exposure durations – often a lot faster than a mechanical shutter can accomplish. However, even if each pixel is exposed for a very fast time, they do not start exposure at the same time, so it takes time to roll through the entire frame, which is what produces the rolling shutter artifacts you see (especially if there is a harmonic relationship between the readout frequency and the motion). This is why it is preferable to use a mechanical shutter
CCD sensors are not made with electronic rolling shutters. They can have a form of analog storage that stores the frame almost instantaneously working as an electronic shutter, which does not produce rolling shutter artifacts (because it is not rolling) but is very expensive (you need 2x the storage cells). As a compromise, a lot of CCDs are interlaced into multiple “fields” and have analog storage for only one of these fields at a time – this allows to use electronic shutter in a reduced-resolution view/video mode which only uses one of these fields, but requires a mechanical shutter in full resolution still capture mode.
The degree of motion blur is directly related to the exposure time (assuming lighting stays the same) at any particular area, a rolling shutter with a very fast exposure time will expose each photo-site without significant motion blur, but since the rolling time is bound by your readout speed, different photo-sites will be exposed at different times. If the motion has some sort of a harmonic frequency relationship with the rolling speed, you can get crazy (but crisp) looking distortion like was shown in the linked pictures.
These sort of effects can also happen with slow mechanical focal plane shutters, in which case it does not matter if the underlying sensor is CCD, CMOS or a piece of film. Like in this picture.