Read Rob's and my messages again. Pay particular attention to words like "significant," "insigificant," and "negligable." A camera is a signal processing system. In such a system, many different components (e.g., lens, film or sensor, physical body, tripod) may significantly influence the quality of the final output. In my engineering experience, it is typical for the limitations of one or two components to dominate. That doesn't mean that performance limitations in the other elements have no effect, just that the effects are insignificant compared to the dominant effects. A negligable difference is not the same as zero difference.
I think that Rob and I are both talking about qualitative estimates, possibly educated guesses, of which component limitations dominate and by how much. You seem to keep beating the "absolute" drum, which isn't what we're talking about at all. The statement "I'm right because 0.999 is smaller than 1.0" just isn't convincing.
If you want to really make your case, you need to define what you mean by resolution. 50% MTF breakpoint in lpmm? Or something else? How do you define detail? And finally, what exactly do you mean by image? Areal image? Image recorded on film? What are the imaging properties of the film itself? etc. Maybe we should use modulation transfer function (MTF) as the measure instead. Are you comfortable with frequency domain analysis? How about linearity vs nonlinearity in signal processing systems? If you make certain special assumptions about the structure of an MTF (for example that it represents a linear filter with a first-order rolloff characteristic), then you can specify its MTF at all frequencies with a single number like the 50% breakpoint in lpmm. If the MTF is more complicated (and they usually are), then the MTF response at a spatial frequency of 100 lpmm doesn't always say very much about the MTF response at, say, 50 lpmm. Then there are issues of accutance vs resolution that I really don't understand very well at all.
I would be really interested to find out if, in practice, the *ist D sensor is the limiting component with, say, a 50mm prime. Norman Koren's MTF measurement methods at http://www.normankoren.com/Tutorials/MTF.html look like a good way to make some measurements....
--Mark
JCO wrote:
I dont understand why you guys cant see that with a given,
fixed, lpmm of resolution, even the very best primes, unless it
is infinity, the cropped APS image will be less detailed than
the FF image. Very simple. And as the sensor's Mp gets bigger,
the difference between FF and APS will get even greater.
The other thing mentioned before is that for a given Mp of sensor,
FF will outperform APS based on larger pixel area and hence lower
noise, greater dynamic range. Bottom line is a FF is better than
APS sensor, both sharper and less noiser, assuming Mp and lens
resolution remains the same.
JCO
