What does the SQF result in the Imatest Documentation mean? I don't understand the graph showing SQF against Picture Height. How should I calculate SQF value for the camera being tested.
For example, the SQF value is 85 when the picture height is 40cm; but the SQF increases to 92 when the picture height is 20cm, and so on. If this is correct, which value should I choose, and why?
SQF is not actually subjective -- it's an objective test with specific criteria, all of them based on MTF.
Now, MTF can be measured in a number of different ways -- for example, the maximum frequency at which a specific level of contrast is maintained. Quite a few sites just quote the numbers for one level of contrast (e.g., 50%, which will typically be quoted as "MTF50").
There is a bit of a problem with that though: there's not one level of contrast or detail that correlates very well with how people will perceive pictures. Just for example, many photographers have argued for years about the differences between the Leica approach (relatively low macro-contrast, emphasis on micro-contrast) and the Zeiss approach (more emphasis on macro-contrast). There's obviously disagreement over which matters most, for what kinds of pictures, etc.
To take that into account, a few decades ago, some people did some studies about what frequencies people perceived as the most important in pictures. They then picked out a set of weights to use for the various MTF tests. Offhand I don't remember the exact numbers, but if memory serves, they average three measurements together -- something like MTF70, MTF50 and MTF30, with MTF70 weighted at 50%, MTF50 at 30% and MTF30 at 20%1. Unless memory fails me even worse than usual at the moment, they also take final print size into account, so at larger print sizes they place more emphasis on finer detail.
Bottom line: Contrary to popular belief, the tests and results themselves are 100% objective, repeatable2, scientific tests. The "subjective" in the name refers to the source they used for the weights, not to the test results themselves.
As to how you'd calculate SQF figures yourself: if you have all the required raw MTF scores, it's a pretty trivial matter of multiplying each by the correct weight, and then averaging the results. The problem is that about the only way to get all the MTF scores you need is to do testing yourself.
1 I'll just re-emphasize the fact that I'm not sure those are the exact numbers they chose. You can look them up if you want, but I don't much care. The point is that it's still an objective, mathematical way of combining objective, scientific tests.
2 Of course, absolutely identical results aren't likely to happen in reality -- there's always some margin for error from differences in equipment, test procedures, test conditions, etc.
Subjective means just that - subjective.
It is an impression based on a persons opinion without an objective measurement. It is how an image is perceived.
Regarding your specific example of higher SQF scores for smaller images: Focussing flaws or negative lens aspects are less visible in a small print than in a large print. So a small print is more forgiving than a large print. This partially relates to the fact that printers have a finite resolution and eyes have a finite resolving power - bluryness below a certain threshold can no longer be perceived without a magnifying glass.
This is why we edit images at 100% when we sharpen - we look at the worst case at large magnification. And what is blurry at 100% might look quite well as a downscaled image or small print.
