Background removal and cleanup of book scan

Dear experts,

i want to use the mighty tools of RawTherapee to tweak my weak scans of old material from archives. Those scans are unfortunately not Raw and they have been taken for a long time also. Mmh, but we will see.

First step is anti-vignetting because at the edge from this library-scanner always some light (resulting in gray areas) was put to the scan. I want to use this filter vignetting and found some positions of the sliders
intensity - 0.44
area 86
0 form
for the left sides of a book.

How to tell the filter that now the virnetting goes from middle to left (because we show a right side of a book)? Sure i can turn the page to 180° but this is much harder to control, because the whole letters are turned up and this is quite desturbing.

Next step i sharpened with
saturation to - 100 (because it will become a bw-publication)
ligtness to +50
contrast + 50
blackworth set to 22000
but the results are not enough. Something in the workflow i am missing.
Can someone help?

A demo can be seen here

(original file)

result is

How to improve my results?

Thanks in advance for any tip,

You can use the wavelet residual layer to smooth the background out:

176-test-mh.jpg.out.pp3 (11.1 KB)

Else in GIMP crop the image to the text, duplicate the layer, gaussian blur until you can’t see any trace of the letters (I used a value of 100 in this example), set the layer mode to divide (screenshot taken at this point), flatten the image, auto-levels and done.


You are a master. I will try!!

Thank you so much!!

I tried to use background substraction in imageJ (sliding paraboloid, 5 px radius). Linear curve adjustment was then done in Gimp.

1 Like

Ups ( i am using the german version and have to look after this function). Where to find it?


Im Englischen ist es “Process” / “Substract Background”. Ich habe hier die englische Version :wink: .

@Pit-Lip this thread will be useful for future reference, do you mind if I change the title to something more easily searchable?

Please change it.
Later, Pit

Changed now to english.
But i can not find it …
Later, Pit

Sorry, seems to be a plugin and not available in the basic package.

Here is another way with only three modules used.
Haze removal, B&W film simulations and the Vignette from the Transform tab with radius set to zero. One can improve the result by using sharpening tool.

90c4162c84d47f41f9abbe5b501166ff29326eb7.jpg.out.pp3 (11.6 KB)

I am impressed shreedhar,

i just put your pp3 on my file and voilá - great!!
Please let me have the right way for control transparent. Where to find Haze removal, B&W film simulations? Vignette i found already …
Thanks a lot!
Later, Pit

1 Like

In RT 5.6, Haze Removal is at the bottom of tools under the Detail tab.
Film simulations are under the Colour tab.
Can not remember if RT comes with film simulations, if it does not, you can download and place unzipped folder under your main RT folder then point to where it is in RT Preferences.

Hazel removal is found, second icon and then bottom. Fine!

Zip is not found. Can you give ressource again?
Thank you very much!
Later, Pit

Using G’MIC. I could make a filter if you want.

This one works
from Film Simulation - RawPedia


very good!! Please, please make a filter - that will make my day!
Thanks in advance!
Later, Pit

Thank you so much!
Later, Peter

Thank you @Pit-Lip. I realised later that not everyone will have a film simulation pack. So here is another try with only two modules. Haze removal and Black-and-White. Both are available with default installation of RawTherapee 5.6.

The trick is to Luminance equaliser mode, increase the Gamma correction to 100% to all three channels and use a very steep `Before’ tone curve with Saturation and Value blending mode. (Seems a mouthful but quite easy!)

90c4162c84d47f41f9abbe5b501166ff29326eb7.jpg.out.pp3 (11.6 KB)

Then one can add the negative Vignette from the Transform tab with the center shifted along the X-axis to fine tune the right side (or left side) to get
90c4162c84d47f41f9abbe5b501166ff29326eb7-1.jpg.out.pp3 (11.6 KB)