Posted February 02, 2010
To illustrate my fascination with science and weird thought experiments, here's something I thought of today. I've wondered about this question before, but today I actually took the time to work out a practical experiment intended to find the answer.
The question is: What is the resolution of the human eye?
In order to look for an answer, we first need to clarify exactly what we mean by "resolution". After all, the human retina isn't rectangular, and ordinary definitions of dot pitch don't really apply. As I see it, in this context the answer can be expressed in two ways. Given that we define a pixel as the smallest distinct visual unit the eye can perceive, the answer can be expressed as either the number of pixels over the diameter of the field of vision, ppØ (Ø being the mathematical symbol for diameter), or as the number of pixels across one degree of the field of vision, pp°.
First we need to establish how wide the FOV actually is. This is done by having the test subject stare at a fixed point straight ahead, then place a distinctive object in his FOV and move it sideways until it disappears from view (remember to move it to the side away from the subject's nose). Simple trigonometry will get you the angle between the object and the line the subject is looking along.
Next, we place a white circle of known diameter against a black background and place it directly in front of the subject (the attached drawing is deliberately misscaled to make drawing the setup at all possible). We now move the circle and background directly away from the subject until he reports that he can no longer see the white circle (if the Earth's curvature prevents this, use a smalle circle). Slowly move the setup towards the subject until he reports that he can see the circle again. Again using trigonometry, knowing the radius of the white circle will get you the arc of vision it takes up at that distance. For the purposes of the experiment we use the radius of the circle rather than the diameter, as well as only the outer half of the FOV in order to eliminate any inconsistencies introduced by the influence of the subject's nose on his FOV.
Now all that remains is to calculate the actual resolution values:
(Av / Ap) * 2 = X ppØ
X / (Av * 2) = Y pp°
And finally, the question is answered satisfactorily. This information will be useful to future designers of visual technologies such as monitors, head mounted displays, AR overlay contact lenses, etc. After all, there is little point in making pixels smaller than the human eye can perceive.
Having thought all this out, I now turn to Wikipedia (doing it beforehand would be cheating, and would also take all the fun out of the exercise), and as it turns out, others before me have studied this in great depth. [URL=http://en.wikipedia.org/wiki/Eye#Visual_acuity]Here[/URL] you will find an explanation of the subject. I must say though, that my experiment attempts to determine another kind of resolution than the one described there. Rather than distinguishing black and white stripes as individual stripes as opposed to a field of grey, I would like to know how wide (in angles) a single white stripe can be destinguished against a black background. In other words, I'm asking not how small objects you can tell apart when they're right next to each other, but how small objects you can see at all.
[/geek] :-D
The question is: What is the resolution of the human eye?
In order to look for an answer, we first need to clarify exactly what we mean by "resolution". After all, the human retina isn't rectangular, and ordinary definitions of dot pitch don't really apply. As I see it, in this context the answer can be expressed in two ways. Given that we define a pixel as the smallest distinct visual unit the eye can perceive, the answer can be expressed as either the number of pixels over the diameter of the field of vision, ppØ (Ø being the mathematical symbol for diameter), or as the number of pixels across one degree of the field of vision, pp°.
First we need to establish how wide the FOV actually is. This is done by having the test subject stare at a fixed point straight ahead, then place a distinctive object in his FOV and move it sideways until it disappears from view (remember to move it to the side away from the subject's nose). Simple trigonometry will get you the angle between the object and the line the subject is looking along.
Next, we place a white circle of known diameter against a black background and place it directly in front of the subject (the attached drawing is deliberately misscaled to make drawing the setup at all possible). We now move the circle and background directly away from the subject until he reports that he can no longer see the white circle (if the Earth's curvature prevents this, use a smalle circle). Slowly move the setup towards the subject until he reports that he can see the circle again. Again using trigonometry, knowing the radius of the white circle will get you the arc of vision it takes up at that distance. For the purposes of the experiment we use the radius of the circle rather than the diameter, as well as only the outer half of the FOV in order to eliminate any inconsistencies introduced by the influence of the subject's nose on his FOV.
Now all that remains is to calculate the actual resolution values:
(Av / Ap) * 2 = X ppØ
X / (Av * 2) = Y pp°
And finally, the question is answered satisfactorily. This information will be useful to future designers of visual technologies such as monitors, head mounted displays, AR overlay contact lenses, etc. After all, there is little point in making pixels smaller than the human eye can perceive.
Having thought all this out, I now turn to Wikipedia (doing it beforehand would be cheating, and would also take all the fun out of the exercise), and as it turns out, others before me have studied this in great depth. [URL=http://en.wikipedia.org/wiki/Eye#Visual_acuity]Here[/URL] you will find an explanation of the subject. I must say though, that my experiment attempts to determine another kind of resolution than the one described there. Rather than distinguishing black and white stripes as individual stripes as opposed to a field of grey, I would like to know how wide (in angles) a single white stripe can be destinguished against a black background. In other words, I'm asking not how small objects you can tell apart when they're right next to each other, but how small objects you can see at all.
[/geek] :-D