Examining 100% crops is not "looking at actual photographs".
Examining 100% crops IS looking at actual photographs, particularly
if the region one intends to use is entirely within the crop. This is
a perfectly valid way to look at things;
It's only valid if you understand what you're looking at.
Some people make the same mistake with MTF. They compare MTF of measured at two totally different spatial frequencies. They'll see lens A has 50% and think it's better than lens B which only has 10%. But what they failed to realize is that lens B was measured at 100 lp/mm, a much higher spatial frequency than the 10 lp/mm used for lens A. If they had measured lens B at 10 lp/mm, they would have found 50% MTF, just the same as lens A.
Of course, you don't see that mistake very often, because it's so obvious that MTF is a function of spatial frequency. But for whatever reason, many people don't realize that the same exact thing applies to image sensors. It's nonsensical to measure things at two totally different levels of detail and then compare them as if they were equal.
How about a car analogy? 100% crop is like driving at 100% speed. Car A has a top speed of 80 MPH (8 MP). The noise at 100% speed (80 MPH) is pretty quiet; not too much wind or engine noise. Car B has a top speed of 150 MPH (15 MP). The noise at 100% speed (150 MPH) is much worse. Very loud and noisy.
Here is where a 100% cropper would say that Car B is worse and start complaining that they don't need 150 MPH and wish the manufacturer had applied their improvements to a car with just 80 MPH top speed, because then they could finally enjoy the peace and quiet they want. But they never even considered that they measured completely different speed. If they had driven Car B at 80 MPH, they would have found that it already
is even quieter than Car A. But they cannot fathom anything other than driving at top speed, and would rather choose a car with slower top speed than just make the conscious choice to drive a speed slower than top speed.
In the same way, NEX Creative did not realize that he was measuring the Canon 50D at a much higher spatial frequency than his 20D.
If you always drive your cars at 100% speed, that's fine, but you have to
understand that 100% speed on one car is faster than 100% speed on another car. Just as 100% crop on one camera is a different image than 100% crop on another camera. To make any comparison requires an understanding of the spatial frequency under examination.
just as, if you're going to use the entire image,
It doesn't matter whether you use the entire image or crop to a certain portion. The point is that 100% crop is a
different portion of the image (different spatial frequency or level of detail) depending on the pixel size.
The world is not comprised of only your viewpoint. You would do well
to open your eyes to other uses, if only so that you can understand
what other people are saying.
I used to be one of the other people. I learned better. Now I'm trying to help the others understand their mistake.
Claiming that the only way that one can look at any detail in
an image is all details at once
That's not what I'm claiming.
, or only when converted to print (a
rarer and rarer occurrence, btw... print media is currently
undergoing a rapid loss of consumers) is 100% wrong.
Printing isn't necessary either. All that needs to happen is for images to be compared at the same level of detail. A crop of a face? Fine. A crop of the center 10%? Fine. But cropping ten faces from a large-pixel camera vs. one face from a small pixel camera makes no sense. Yet that's what 100% crop does.
scaling is not limited to
the domain of the print, or to the entire image.
Agreed, of course.
...and if YOU looked at images in real life usage (a crop of a nebula
out of the night sky for example,
Cropping a nebula is fine. the 100% crop comparison method applied to a nebula:
large pixels: crop out the whole Orion Molecular Cloud Complex.
medium pixels: crop out just M42
small pixels: crop out just the Trapezium.
That makes no sense. It's comparing completely different compositions. The correct method is to crop out the
same portion of the image no matter what the size of the pixel. If you're shooting M42, crop out M42 from every camera. Shooting just the Trapezium? Then crop just that out. But don't apply different crops to each camera, thinking they are the same just because they're all "100%".
or a person's face clipped out of a crowd,
Again, that's fine. But that's not what 100% crop is about. 100% crop on a 3 MP camera gives you 10 faces. 100% crop on a 12 MP camera gives you 5 faces. 100% crop on a 48 MP camera gives you 2 faces. I'm saying that a comparison must be based on cropping the same no matter what the pixel size.
or a single performer clipped out of a band,
or one car clipped out of a field of racers,
or your kid clipped out of the school play,
Those are all great ideas, but still have nothing to do with 100% crop.
What you ARE wrong about is presuming that your way is the
only way.
Anything that doesn't account for spatial frequency is wrong. People are wrong sometimes, and they can continue to persist with their incorrect understanding, but
This is the real world. Not the "print only" subset of reality you
talk about,
In the real world, it's a mistake to apply different standards arbitrarily. Only when the same standard is applied to everything under comparison can a conclusion be drawn.
--
Daniel
