On Tue, 29 Jul 2003 18:40:21 GMT, (Don) wrote:
I found a very clever and revealing resolution test which comes at it from a slightly different angle. It’s based on the fact that interpolation does not create any new detail. The test comprises scanning the same image twice, once at high resolution (say, 2400) and once at lower resolution (say, 600). The low resolution scan is then interpolated, or should that be extrapolated? 😉 to 2400 dpi.
Putting these two images side by side (now of identical sizes and so much easier to compare), examining them closely and even going down to pixel level will show that the genuine 2400 dpi scan does not reveal any more detail than the "make believe" pixels "invented" by the interpolation process in the other image! Game, set and match!
Don.
BUT I have two rhetorical a question about that…
1. Is the judgment just a visual inspection or did you actually measure the color value of pixels in the same position in each scan? The reason to ask this is that a greyscale image pixel can be any of 256 colors and subtle changes would mean that your 2400 dpi scan would still be more accurate (though how significant the difference I’ll leave up to statisticians).
2. I wouldn’t expect that the majority of pixels would be different in the two scans cited above. After all, adjacent pixels in, for example, the white area in a cloud would not be that different in the native 2400 dpi or in the 600 to 2400 dpi conversion. The real differences only come into play at the edges where the pixels are changing color to convey a new structure (detail).
AND NOW… Just as a complete aside, here’s another completely different image size conundrum:
We often publish photomicrographs and electron microscopy images. The authors typically submit them as 5×7 prints and in the captions to the photos they specify the microscopic magnification (i.e., X200, X400, etc.).
We’ve has a few authors that insist that if we resize the image from 5×7 to 3×5 then the magnification in the photo’s caption should change in the same proportion.
In an explanation to one author of why not to do that I gave the example of viewing the 5×7 image from 2 feet away then from 4 feet away (would the magification change?). How about if the image had been submitted as an 8×10, would they have increased the magification used on the microscope?
Then when you layer the fact that the image is going to be printed with a 150 halftone line screen, some authors seem to get really confused about exactly what resolution that the microscopes lens is conveying.
— JC