One important message we cannot give enough is "pixel count is not resolution". We see resolution as the capability of an imaging system to reproduce spatial frequencies. The higher the frequencies that can be transferred into image content, the higher the resolution.
The complete behavior of the device under test is described by the spatial frequency response (SFR) or the modulation transfer function (MTF). To make a comparison easier, these functions are often reduced to single numerical values. For example this can be the MTF50 value, which is the frequency that leads to a modulation transfer of 50%. The MTF50 value is more related to the subjective "sharpness" of an image, while the lens and system performance can be seen better in the MTF10 value, also called "limiting resolution".
|
| The MTF plot for the Leica M. The limiting resolution and the MTF50 value can be calculated. |
So to obtain numbers like MTF50 of MTF10, the MTF is calculated first and then these numbers are derived from it. Now an important problem in automated and mass testing which we face in the iQ-Analyzer. What do we report if these numbers cannot be derived?
|
|
|
The Leica M Monochrome has no Bayer pattern, which results in a higher modulation transfer for all frequencies. A MTF10 value can not be calculated below the aliasing region. The results are marked with a "!" |
A lot of engineers use the iQ-Analyzer in their lab for bulk tests, so they test a huge amount of cameras or lenses in a short period of time without user interaction. So it is no option to raise a dialog in the software and to ask the user what to do.
We decided to use the following:
If the iQ-Analyzer can not calculate the limiting resolution (this is where this problem most likely occurs), it will use the Nyquist frequency instead. For example for a system with 3000 pixel in height, the reported number equals 1500 LP/PH. But how to differentiate between a really measured 1500 LP/PH and the set value of 1500 LP/PH? In the user interface we report the set value as "!1500LP/PH". So from the leading "!" you can tell, that this value can not be measured and was set to the theoretical limit. In the stored result files, the value is written as "-1500", as this makes the further processing easier, it can be read in as a number.
Do you have a better idea? Let us know!
Best regards
Uwe Artmann