|
Hello everyone, I'm trying to address an issue here that is often misunderstood by the vast majority of amateur photographers, ie the resolution.
I apologize in advance for any mistakes I might say. That is the risk you take when you put yourself in the position of "teacher". So any comment, criticism, objections are welcome.
The common idea is to think that the more pixels there will be more determined and our image.
Actually there are two elements that will contribute to the resolution of an image: The diameter of the diaphragm opening, and the pixel size.
I'll start by giving some basic optics, so that everyone may appreciate the physical concepts behind these terms fairly airtight.
RESOLUTION
The resolution is commonly a concept that allows us to say we see more details. And the resolution is characterized by the "separation" between two points of light from which we distinguish these two points on camera.
For example, a person who has poor eyesight and two nearby stars look (Alcor and Mizar in the Big Dipper, for example) will do that. While a person may be less easily distinguished old 2 stars. The two stars were also cited the visual acuity test that allowed in the Arab world of the Middle Ages, to determine the quality of the sight of people.
Our eye operating in the same way as a camera (at least we have a pupil who grows and shrinks like a diaphragm), we can draw a parallel with the photograph.
So the resolution of a camera is characterized by its power of separation.
This separation is expressed in angle. Why? It's simple:
Imagine that you were on the beach. You plant two sticks on the sand and wait until you pass a boat. The distance between the boat when it is masked by the first stick, and when it is masked by the second is even bigger than the boat is away from you.
However the angle is the same. So on a picture, a distance between two objects is expressed at an angle.
Since our diaphragm does not have an infinite dimension, when we made the image of a point (a star far away for example), we get a point but not a spot (the Airy disk). That's DIFFRACTION. This task is even larger than the diaphragm is small. Conversely, if the diaphragm was infinitely large, we would not have a job but a point.
This is very important for the resolution. Indeed, consider two points very far as the headlights of a car in sight.
By the image of the car (at night), the photo will appear as a single point. Why? Because of the 2 headlights DIFFRACTION creates our image 2 tasks that are so close they seem to form a kind of big job deformed.
The resolving power of the camera is not big enough.
What would he have had to do to distinguish 2 points on the photos? He would have had to reduce the diffraction effect, enlarge the size of the target of several decimeters or even meters.
In this case, the optics of the camera would have the resolving power needed to distinguish the two lighthouses.
But another problem arises and this brings us to the second chapter: the Pixel.
THE PIXEL
The pixel of the CCD during the grain of the film with film. Our CCD consists of a grid of pixels, just like alcoves hives of bees, except that our pixels are square. Each pixel receives a certain number of photons it "converts" into a number of electrons. This number of electrons is recorded for each pixel and this is what shapes our image on the computer. The conversion quality is called noise.
Indeed, if a pixel receives 2 photons ... sometimes it is wrong and converts them into 3 electrons. The error is 1. This is especially visible when you go up the ISO, where each conversion is multiplied. ISO 3200 and a pixel must convert photons into two 32 electrons (which can see better at night when there is little light, few photons) if the pixel is mistaken and has 3 photons instead of 2, it is the converted into 48 electrons! The error is 16!!
This is mainly related to the electronic quality of the CCD.
Returning to our problem resolution. We have seen that the diaphragm of our device creates a stain instead of a point. This effect is obviously harmful to the quality of the image.
Fortunately, if this task is smaller than the pixel size of CCD, voila. Your CCD will count the number of photons arriving on the pixel and will turn it into a number. It remains below a critical threshold where task
Unfortunately, if you close your diaphragm (say f/22) the task will expand even more than the diaphragm is small. What's going on?
The task will start to overflow of the pixel and neighboring pixels illuminate. So you will save several CCD pixels for the same information ... It has surpassed the fateful task> pixel
This reasoning also applies in reverse.
Resume our normal opening, f/1.8 for example. We can even take f / 1 the problem is the same. Our objective this time will create a bright spot, a small task!. Once you're happy.
But if we reduce our SIZE pixel too, they may become so small that their size is smaller than our task! pixel pixel
That's what happens when manufacturers increase the number of pixels on the new models: Consider a puzzle of 500 pieces. If you want the same puzzle the same size (to put in the same context, for example), but with 2000 pieces, you understand quickly that the new puzzle pieces are much smaller.
It is with our CCD. Their size does not change (24x36 full-frame for example). But the race for the number of pixels is actually a race to shrink the size of the pixel, until our fateful threshold ....
So there would be interest in keeping large pixels (and suddenly have less) which would always remain above the threshold. In addition an element plays in the favor: the noise mentioned above. Imagine the same size CCD. The first of which 1 million pixels, and another 4 million. So where the first pixel CCD can accommodate 1, the second can accommodate 4. Now a pixel to 4 times less likely to be mistaken in his conversion photon-electron only 4 pixels ...
All this encourages us to increase the size of our pixels and also decrease the number.
But then another problem arises. That of sampling, and this will be my third and final chapter:
SAMPLING.
The sampling frequency which is transmitted information.
When you watch television, the refresh rate above 25 frames per second (refresh rate of the eye) which gives the impression of movement and continuity. However when the reception is worse, then the sampling rate falls below 25 frames per second and it gives an impression of jerky that is noticed
For our image, same thing. If there were only 4 pixel per picture, we would see 4 large squares (eg, 1 white, 1 green, 1 red and one blue) and that's it. Gradually, as we increased our sampling of the image, we increase our number of pixels for the image by reducing their size, then the image becomes finer and finer details appear.
Another example, turn on our beach, and put buoys in the phosphorescent water, straight out to sea. Imagine that we put a whole 50 meters. When night comes, we only see the buoys. Because they are very distant from each other, their movement does not allow us to guess the waves. At worst, they can all move in unison if unfortunately their inter-distance is exactly the distance between two troughs of the waves ... Now retentons experience and place a buoy every 5 centimeters. The night arrives and the beautiful line described by the buoy vertical movement of each of them, the characteristics of waves that raise.
Our CCD is identical, the buoys are the pixel and the waves are our image. We seek to see the image that is the wave through the buoys. So we want a lot of buoys.
Thus, in the interest of making a beautiful picture detail well, we are tempted to add a large number of pixels.
But we saw in Chapter 2, this increase to a limit: the fateful threshold.
If it rises too the number of pixels, their size will decrease too (because the total size of the CCD remains unchanged as our puzzle) and we will pass the critical threshold: pixel
DISCUSSION
We find that the technical quality of an image is a clever balance between pixel size, pixel count, and quality of optics.
Fortunately the film has been there and we empirically given references. The 24x36 is the "norm" for the dimensions of the CCD. Close an f/22 lens gives practically the fateful threshold for the size of a pixel. And thus a maximum number of pixels not to see diffraction. Coincidentally the best compromise is the grain size of silver films
And do not forget that each image is composed of the superposition of our small spots due to diffraction.
Even when I do the photo of my brother in red sweater, each photon arriving from the red sweater to my goal will be transformed into a task. And all the red spots will pass through a sieve, my CCD which transform all these tasks in light-mail tasks, making few mistakes ...
Related articls :
|
