1
\$\begingroup\$

I have scanned this photo from a book that's printed with color ink, but the picture is a black and white one. When I scanned it with Vuescan as a color photograph I get vertical lines like

enter image description here

I think it's called a Moire pattern, but I haven't seen one like this one. How can i prevent them or how can I get rid of them in Photoshop or other software?

\$\endgroup\$
4
  • \$\begingroup\$ Have you tried scanning as B&W? \$\endgroup\$
    – MikeW
    Oct 16, 2015 at 22:05
  • \$\begingroup\$ Try using a diferent scanning resolution or try taking a picture. \$\endgroup\$
    – Rafael
    Oct 17, 2015 at 1:40
  • \$\begingroup\$ Unfortunately I don't have the original book \$\endgroup\$
    – PerseP
    Oct 17, 2015 at 19:59
  • \$\begingroup\$ I had the same thing happen to me when I scanned film negatives. In the case of film negatives, I think it's even weirder (and not solved on this page) \$\endgroup\$ Oct 18, 2015 at 1:13

2 Answers 2

1
\$\begingroup\$

This is aliasing, often called a moire pattern when it occurs against something regular, as in this case.

In this particular case, the actual picture is made of lots of little dots of ink in a regular pattern. Your camera sensor is also lots of little sample points on a regular pattern. When the spacing of the two are close (or multiples are close) but not the same, you get this moire pattern or beat frequency.

Scanning ink-printed halftone pictures is tricky. The best way is to scan at such a high resolution that there are a reasonable number of pixels per ink dot. Or put another way, use enough resolution to resolve each ink dot separately. Then you can digitally filter the image down to a more manageable resolution. Proper digital filtering will cause the ink dots to be averaged out from the initial high res samples, not just re-sampled at a lower resolution.

Think of a black and white ink halftone picture. You scan at high resolution, so you can see the individual ink dots in the scan, and how big each one is. In theory, all the pixels should be either full black or full white. In fact, it often helps to amplify the range about some midpoint to get mostly black and white pixels, since that's what the print actually has. At a fine detail as your scan, the print is either white where there is not ink, or black where there is ink. This may take 1200 DPI or more to obtain without noticeable moire pattern.

Now you can shrink this image to whatever resolution you really wanted, like 300 DPI for example. The digital filtering and shrinking process will take a bunch of original pixels into account to get each output pixels. This allows the output pixels to be shades of gray.

\$\endgroup\$
1
\$\begingroup\$

Yes, in common parlance what you're seeing is a moire effect as that is used as shorthand for this type of interference pattern, you could also call it aliasing. Certainly if you used the term moire in relation to the image above photographers would understand.

The reason is that, to get the image to print, the printers have used a technique called halftone. In this process an image is turned into dots of differing sizes and/or spacing that our eyes generally don't see as the dots are quite small.

Being able to control/fix it really depends on what original scans/sources you have, but the general process is: starting with a high resolution image in which the dots can be seen individually (I'd go for minimum 600dpi scan), you apply an appropriate amount of blur to turn the dots back into graduated tones and then resample the image to reduce the DPI to a more reasonable level.

Like this quick & dirty simulation of the process in this cropped image...

Colour halftone image created in Photoshop
An initial halftone image.

Colour halftone image which has been pixelated
The same image pixelated to represent a low-res scan.

Colour halftone image which has been blurred then pixelated
The same image pixelated after a Gaussian Blur applied. It still needs some clean-up but it's a great improvement.

Without a high-res starting image, you're pretty much done for as the interference is essentially 'baked-in' to the image you're working with.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.