Actually there’s some misunderstanding here.
An image capture device like a phone or digital camera, etc. captures a fixed amount of data - 8MP on an iPhone 6s. This means is captures 8,000,000 (rounded value) pixels - no more, no less.
Here is an image I just shot and Synced to DTPO…
2448 * 3264 = 7990272 (~8,000,000MP)
But what about the 72dpi? That’s a relative value. Relative to what? Relative to the requirements of the output or process. Follow this…
2448 / 72 = 34
3264 / 72 = 45.3
Ok, so what? Those values are inches! Remember, dpi = dots per INCH. So you could output this image at 72dpi at 34x45 inches. That’s the amount of data in it.
Now, outputting at 72dpi wouldn’t be a good quality product, but unless you have a large format printer, you’re not doing such a thing. (On a potentially interesting side note: 72 dpi at large sizes are used for billboards as: (1) higher resolutions would make for terabyte-sized images, and (2) the viewing distance is so great you don’t notice any quality issues!)
Now, let’s look at the math again, but using a value appropriate to commercial printing and OCR: 300dpi.
2448/300 = 8.16
3264/300 = 10.88
So there’s enough data in the image to print or OCR a(n approximately) US Letter size document.
By contrast, here’s a marqueed selection from a non-Retina Mac…
331/72 = 4.6
296/72 = 4.1
So you could use this screencapture at 72dpi at 4.6x4.1 inches. Good for onscreen, passable for a quick print at home, bad for commercial printing.
331/72 = 1.1
296/300 = .98
So you’d get a little over a 1x1 inch image at 300dpi.
Now if you scan a US Letter page at 72dpi, you’d have pixel values of 612x792. Less than half a megapixel of data. This has insufficient data for anything but onscreen viewing.
Scan it at 200dpi = 1700*2200 = 3.4MP.
And at 300dpi, 2550x3300, 8.4MP. This is sufficient for OCR as well as commercial printing.
So, blah blah blah, what does this mean… it means there is actually sufficient data for the process. Granted, the higher the megapixels of the capture device and the quality of the lens will factor into things, but you should be able to process the image.
Here is a PDF at 150dpi and 75% Quality (264.7KB)…
And one at 300dpi and 100% Quality (5MB - consider that)…
Not an appreciable difference and NO, you shouldn’t be looking at them this closely to determine quality. You don’t read a PDF zoomed in this closely.
(And to head off the obvious question: No, scanning at 600 or 1200, etc. isn’t better. If you have a document with a lot of fine or thin detail, 600 could help but more data in other cases is increasing file size with no real benefit.)
On a side note:
Upsampling increases the megapixel value.
Downsampling decreases the megapixel value.