A few scanning tips

This is a 6x4 inch print scanned at 60 dpi. It is shown for the purpose of establishing scale for the following images taken from it. Just an ordinary snapshot, but with a lot detail. The film is 35 mm Kodak 200 Gold. The film processing was 1-hour by WalMart, paper was Fuji. Camera lens is a Nikon 50 mm f/1.8 AF-Nikkor (not a zoom lens). The subject distance from camera tripod to that thermometer is about 30 feet. All images have been sharpened slightly, but all the same.

Here are sections of scans from this print at 100 dpi, 200 dpi, 300 dpi, 400 dpi on a 600 dpi Microtek ScanMaker III.

Allow me to suggest the only viable way to make the comparison of two scanning resolutions.

On the left below is the 100 dpi scanned sample, but resampled to 2X in Photoshop (bicubic), for the purpose of comparing to the 200 dpi scan, repeated here on the right. A 2X upsample does not generate the best image, but it is the only way to compare detail with the larger 200 dpi scan. We cannot compare different sizes, the brain is too easily fooled, and big wins every time. If we just want "big", say for the video screen, then the higher resolution scan absolutely delivers "big". But for printing, the only reason we want big is for detail, and the point of this exercise is to see if we get detail. We cannot resample the larger image down, because that would discard the detail that we hope to compare. Resampling the small one up is the only way, however we know this introduces problems. The only meaningful result is if it holds its own in spite of this.

There is noticeable difference in detail between the two of these. The problem might be blamed on the upsample of the smaller, but it is very easy to believe that 200 dpi scans of a color print do show more detail than 100 dpi scans.

OK, the difference here is WHAT we are looking for.

Let's look for it.

On the left below is the same 200 dpi image before, but now resampled to 2X (in Photoshop, bicubic), for the purpose of comparing to a 400 dpi scan, on the right.

The difference in image detail between the two is nonexistent. Sure, the 2x upsampling does add some jaggies, the interpolation artifacts are not as smooth as the higher resolution scan, but no item of actual image detail shows better in one than the other. Which suggests the 400 dpi scan didn't add anything. There's no point in quibbling about how the 2X upsample is not optimum, because the straight 400 dpi scan is no better. They are essentially the same image. We really don't gain additional detail from scanning color prints over about 300 dpi.

However, scaling to print the smaller image larger also gains size without additional detail, and without interpolation. For printing, the smaller 300 dpi file will normally do as much as the larger 400 dpi file will do, simply because this color print did not contain more detail itself.

But there are other valid opinions, not about the detail, but about the upsampling. You may like the 400 dpi image better than the 2X resampled 200 dpi image for reasons other than detail. Upsampling the 200 dpi image to be 2X larger above caused its jaggies to be more visible. Images at 200 dpi and then scaled 2X larger do have more jaggies than if scanned at 400 dpi. Video screens show every image pixel directly, where printers normally show a more averaged view.

Therefore, my previous suggestion was, if we need to scan color prints for large images for printing, then scan at 300 dpi and hope it helps somehow. It can help some, and it will remove any uneasiness that more scan resolution might help, and will add the warm fuzzy feeling that we have done all we can. If you specifically want a large image, there's nothing wrong with scanning at 600 dpi, unless memory or disk space is a problem. But if a problem, scanning color prints at 300 dpi will be as effective.

For printing purposes, scanning color prints above 300 dpi won't help much, but won't hurt either. However, you will pay a big price in bytes for image size. That may be no problem for a big computer, and when scanning film, it is clearly justified. But I hope you understand why you won't get the detail you might hope for from prints over 300 dpi.

On the left below is the same original 400 dpi scan of the color print, and on the right is a 1950 dpi scan of the negative on a 30 bit Microtek 35t Plus film scanner.

Both images have had the same sharpening treatment. Nothing is faked. There is a considerable difference in the detail achieved in the same size image. Remember, we're 30 feet away with a 50 mm lens, and this is 3.4% of the total film area.

That these two are far from equal is the best proof that prints cannot deliver expected detail over 300 dpi. Film is better.

Megabytes are very desirable when printing large images.

Newer scans of the same negative with a Minolta Scan Dual II film scanner at 2820 dpi, and with a Polaroid SprintScan 4000 at 4000 dpi, are HERE.