Photoshop Wishes

Suggestions for Adobe Photoshop

FFT-based pattern remover

UPDATE:  

New filters/actions for doing FT-based pattern removal can found HERE.
Go there instead of following the tutorial in this post. The process has been greatly improved and semi-automated since I made this post.


Some applications offer FFT/IFFT-filters that can be used in a tedious process to remove repeating patterns (like raster patterns). I wish Photoshop would become the first application to automate this process.

It would work like a filter that – as if by magic – removes repeating patterns in the image:

A quick sample of what can be achieved with the help of FFT. Here a paper texture is removed.

This is done by performing a “Fast Fourier Transform” (FFT) of the image, splitting it into frequency amplitude and phase (storing it as separate channels). Repeating patterns (frequencies) stand out clearly and can be suppressed by the filter. An Inverse FFT is then performed to convert back to a normal image (now with the repeating patterns removed or reduced).

Note that this only works on a single channel, so normally you would just bother to do this on the Luminosity component. Filtering each RGB-channel separately should be considered. I will leave that up to you Adobe… 😉

For the user, this could be a simple one-click operation, or better yet a slider that affects how aggressively the filter removes frequencies.

Read more to see how to do this in Photoshop manually step by step.

How to remove repeating patterns using FFT

First, download and install the FFT/IFFT-filters (Windows only). Follow the readme.
The filters were originally made by Alex Chirikov, and updated by Phil Thornton.

Then we start off with an image that suffers from an evenly repeating pattern (like this paper texture or a raster pattern):

(Images have been cropped and reduced in size)

We then run the FFT-filter and end up with a whole lot of seemingly random noise.

The interesting part is the red channel that shows the amplitude of our frequencies:

In a normal photograph we would expect only to see the bright star in the middle, but here we also have lots of small bright star-shapes scattered across our frequencies. Those little stars make up the strong repeating patterns (frequencies) in the image

We remove them using a black brush. This part is tedious but I believe it can be automated and thus making the whole process one simple operation for the user. You have no preview of the result while doing this, so just cover up as much as you can:

When done, run the IFFT-filter to see the result. You might have to go back and forth a couple of times:

I call this amazing! I almost never encounter such images, but I still want this as a filter!

The output is grayscale, so for color images blend the result with the original using Luminosity blend mode (or perhaps run it once per channel?).

Crop of a poor scan from and old newspaper.
Raster removed using the described FFT-method, and Healing Brush used to repair damaged parts.

I have submitted this to Adobe.

Category: Filter, Major change
  • Dean Messer says:

    While trying to find a FFT file I have encountered every genus wanting to prove their own intellect. All I want and need is a safe working copy of FFT to use with Windows 7 (hopefully 64 bit) and instructions how to unzip and where to put it in what adobe folders.

    With that said, your site looks more professional than most I have encountered. I downloaded and saved the file in my down load folder and I have WinRAR as my unzip program.

    I have used and liked Adobe for many years, but at my age I don’t want to deal with long learning curves from one program to another. Adobe seems to think making things complicated either boasters their collective powers or justifies their high priced products. A ffT plugin would be simple enough to incorporate in their design and there are other plugins like, Color Wash, Contrast Master, etc. that corrects all sorts of old photo conditions without going through a gross number of labor intensive steps.

    Regards
    Dean Messer

    November 29, 2012 at 5:16 am
    • Jonas Madsen Rogne says:

      I have tested the plugin I linked here and I can confirm that it works well with Win 7 64-bit. It is the only FFT-filter I’ve found for 64-bit Photoshop (Windows only…).

      This is the forum thread where I originaly have it from:
      http://www.retouchpro.com/forums/software/33084-here-fft-64-bit-photoshop-2.html

      There is a readme-file included with install instructions (just place the plugin-files in your plugin folder):
      * Unzip/unpack
      * Copy the two plugins (.8bf) to the plugin-folder:
      “C:\Program Files\Adobe\Adobe Photoshop CS6 (64 Bit)\Plug-ins”

      November 29, 2012 at 8:21 pm
      • Tim O'Connor says:

        Alas this does not appear to work with CS6 – or at least with CC.

        June 27, 2013 at 3:38 pm
        • Jonas Madsen Rogne says:

          It most definately work with CS6, but I just tested with a trial of CC, but couldn’t get it to work (I only tested 64-bit).

          June 27, 2013 at 7:55 pm
          • Jonas says:

            Strange, now it works in CC. Perhaps there was a CC update?

            August 13, 2013 at 2:53 pm
  • Boyd Murray says:

    I have tried your ‘FFT-based pattern remover’ technique on a colour photo. After the IFFT operation, the repeating patterns are somewhat reduced as required; but my original color photo comes out as black & white.

    I would be most appreciative if you could tell me the ‘FFT-based pattern remover’ technique for a **COLOR** photo.

    Best Regards …

    June 27, 2013 at 8:55 am
    • Jonas Madsen Rogne says:

      The output is Grayscale no matter what, so for color images blend the result with the original using Luminosity blend mode. You might need to blur the colored layer a bit to smooth out any remaining color pattern.

      Perhaps run it once per channel? <- I haven't tried this because it's too much work...

      June 27, 2013 at 7:57 pm
  • Wayne Kelley says:

    I have searched and gone to all the links and to no avail, I cannot locate a 64bit version of FFT for CS5. The links come up 404 not found.
    Any suggestions?
    Cheers,
    W. Kelley 13/12/02

    December 3, 2013 at 4:19 am
  • theo kats says:

    I’ve tried both CC 32 and CC 64. In both the plugin appears to work howver when I run the FFT filter I get a red noise screen with no ‘stars’. Checked my red channel and it’s pure white..

    May 7, 2014 at 8:11 am
    • Jonas Madsen Rogne says:

      This will happen if the image is too large unfortunately. One of the things I wish was fixed (along with OSX support).

      It’s hard to tell exactly where the limit is however, but you will have to downsample your image some before running the filter if it fails. :/

      May 7, 2014 at 1:24 pm
      • Ang says:

        Hi, I was trying this for my school project and encountered the same problem as you. I tried downsizing it till it was around 48.1 MB before it starts working properly, so I can assume that 48.1 MB is its maximum file size limit.

        The would probably limit your ability to print at larger sizes though, which can be an issue depending on your requirements (my prof wanted 64MB for A4 printing).

        Correct me if I’m wrong, but since these filters merges and converts the image to black and white; it is probably advisable that you make adjustments to your colour channels etc. for clean up purposes prior to performing this texture removal.

        February 7, 2015 at 8:23 am
        • Jonas Madsen Rogne says:

          You can do other cleanup on the image both before and after. 🙂
          I would perhaps separate out the luminosity and run fft on that, then blend it back.

          You could test using fft per channel but i suspect you will have to make sure the exact same edit is done to each channel.

          Ps: For small prints that are to be viewed up close you want 300 ppi… The teacher using file size as a “requirement” instead is weird (less accurate and inconvenient).

          February 7, 2015 at 8:38 am
  • Jonas Madsen Rogne says:

    UPDATE

    There are now new FFT-filters for Windows, greatly improved, with actions to semi-automate the removal.

    https://www.retouchpro.com/forum/tools/software/39278-ft-pattern-suppressor

    (A v2 will be out shortly with more improvements, and color/per-channel/3D suppression).

    May 13, 2018 at 2:03 am
    • Victor says:

      That’s a great tutorial, easy to follow. Maybe I’m wrong, but…is it possible that the pattern is so intense that is present to some degree on all the RGB channels? I apply this technique in the red channel, but I see very little or almost no improvement. Thank you 🙂

      May 13, 2018 at 4:40 am
      • Jonas Madsen Rogne says:

        Hi

        Try the plug-ins and actions from here: https://www.retouchpro.com/forum/tools/software/39278-ft-pattern-suppressor
        (There is a video with step-by-step).

        It will help you get much better results, and in less time! 🙂

        The filters/actions at the moment does not take color into account, so you have to blend any colors back in afterwards yourself. We have created an improved v2 that also handles per-channel/color suppression, so if you need that, I uploaded a beta for you to test (no tutorial video recorded yet):
        Download files
        Readme

        Regarding your question, with the basic approach (and old filters) shown on this blog post, then the frequency information is in the red channel and not present on green/blue. The green is the phase information and might show signs of the same patterns, but it is not necessary to do anything there. The blue channel is not used for anything (it’s ignored) and you can paint a smiley there if you wish. 😉
        If you have a challenging image and problems getting good results, it would be great if you could try the new filters/actions at the link I gave you and report back on the retouchpro-forum if you have any problems (and it’s awesome if you can post the image so we can test ourselves – as every image is a bit different and we’ve only had a limited set of images for testing).

        May 13, 2018 at 11:10 am

Your email address will not be published. Required fields are marked *

*