Aliasing Artifacts and Digital Cameras

Most recent answer: 10/22/2007

Q:
How badly, if at all, would you expect the digital camera to generate aliasing artefacts? thanks
- John McEntyre (age 17)
Holmes State School, Australia
A:
Hi John,

Sure, digital cameras can create aliasing artefacts, no problem. Aliasing distortions arise when a signal (be it light or sound or anything else) which has a periodic structure is "sampled" at a frequency which is a poor match for the frequency of the signal. Whenever a signal is converted to a set of digital numbers, these represent the signal at discrete locations (times for sound, positions in an image for a picture). And if these locations are poorly chosen, well, artefacts occur.

Many objects people take pictures of do not have periodic structure, and then a digital camera should not introduce its own aliasing artefacts. You can get artefacts however if the digital picture is then re-sized onto another grid of pixels using a computer program. This is a very common picture distortion.

Going back to your question, however, of whether just the camera alone can make aliasing artefacts, the answer is definitely yes. You can take a picture of some graph paper to see if you can get it to happen. Pick some with high contrast (very light squares and very dark grid lines), point a digital camera at it, and walk slowly away from it (or experiment with the camera’s zoom feature if it has one) so that the grid lines on the graph paper get progressively smaller and smaller in the picture. Be careful to make sure the picture stays in sharp focus all the time. Eventually, you may find a distance/zoom combination which can make the graph paper either all very dark or all very light depending on slight variations in how you hold the camera (you may need a tripod to hold the camera steady enough, but you will probably notice something strange going on anyway). Or you may observe interesting geometric patterns, particularly if the graph paper is not held flat or is held at an angle relative to the camera pixels. Experiment with different orientations, distances, and bend the graph paper, etc..

These effects happen because the places the camera chooses to sample -- where the pixels are -- either may line up with the graph lines or the spaces between them. They have also been observed to happen when taking pictures of polyester cloth, tweed, and herringbone patterns, although sometimes you have to look hard for them.

Note 1: The viewfinders of these cameras are composed of arrays of pixels too, and they may have fewer pixels than the picture is taken with. This introduces aliasing as the main pixel array is sampled a second time on a coarser grid and you may find that the viewfinder image differs from the final picture that gets taken if you are taking a picture of a periodic object.

Note 2: Digital cameras with just one CCD arrange it so that adjacent pixels sample red, green, and blue light separately. This makes aliasing artifacts even worse because your picture of white and black graph paper may have streaks of red, blue, or green in it. Here is an example:

.

Note 3: Some digital cameras may have "anti-alias" filters which blur the incoming optical image to prevent these artefacts, and so if you do the experiments and do not find the artefacts, it could be you have a nice digital camera which tries to prevent this kind of thing.

There is another kind of aliasing which can plague digital photograhers. Not only does a digital camera sample on a grid in space, it also does its sampling over a short period of time (so that pictures of moving objects can be as sharp as possible). The camera repeatedly samples and resamples its view and shows the results on the viewfinder. If something in view is flickering at a frequency which doesn’t match the camera’s update frequency very well, you may observe a timing aliasing effect. An easy experiment to do is just to take a digital picture of a television set or computer monitor. It may look odd in the viewfinder too. This is a problem also with ordinary cameras with shutter speeds faster than 1/60th of a second. See  for more information. Your digital camera’s refresh rate may depend on how much light is available -- experiment! I saw this effect using a digital camera both with a television set and separately also with the propeller of a turboprop airplane on a runway on a bright, sunny day.

Tom

(published on 10/22/2007)