astrolucida
Offline
Upload & Sell: Off
|
 p.5 #3 · DPreview of 50D is flawed ..please read | |
George.ML wrote:
Suppose that you print an 8"x10" photo at resolution 300ppi (pixels per inch) and then you decide to scan it.
The printed photo has a fixed resolution of 7.2mp (8x300x10x300).
If you use a 300ppi scanner, you will recreate the printed photo 1:1 (that is, 100%).
But what happens if you decide to scan it at 600ppi resolution?
With 600ppi, each pixel from the original photo will be represented with 4 pixels in the scanned image.
That is, you are getting a 4x magnified (digitally) version of the original photo.
You are here assuming that the scanned pixels would align perfectly with the printed pixels. However, this is never the case. Instead, one scanned pixel would always sample from several printed pixels, in effect blurring the image. The 600 ppi scan would recover practically almost all information from the print while the 300 ppi scan would actually lose a significant portion of it. And here we are assuming that such a thing as a "printed pixel" actually exists - in reality each "printed pixel" consists of a group of printed dots. This further blurs the issue - but is not significant here, though.
The same thing applies to lenses: only by "oversampling" you can get all the detail that a lens can produce. Naturally, an oversampled image looks blurry when compared to an "undersampled" image.
This is what we are seeing now with the recent camera. Instead of saying that a camera starts to outresolve the lenses, we could say that the lenses still have some resolution remaining that we cannot yet tap with the cameras.
In my experience, a 450D can squeeze almost even the last bit of what the 400f2.8L II lens can provide, but only by stacking the lens with two 2x teleconverters! I have verified that by taking Moon images with the 2x converter, with stacked 2x and 1.4x converters and finally two stacked 2x converters. The actual Moon detail increases at each step, though the difference between the steps becomes smaller and smaller. The resulting image looks soft, but with post processing it can be made to look sharp, even at 100% pixels. Yet there seems to be any more detail from the optics that the camera would miss.
If you want very sharp images, with no post processing, you need to have the lens outresolve the camera by a factor of 2-3. In effect, losing some potential detail. On the other hand, if you want to get all the detail that the lens can provide, you need the camera to outresolve the lens by a factor of 2-3. This is what is done in astrophotography, when you want to maximize the amount of actual subject detail.
|
