Daniel Reetz, the founder of the DIY Book Scanner community, has recently started making videos of prototyping and shop tips. If you are tinkering with a book scanner (or any other project) in your home shop, these tips will come in handy. https://www.youtube.com/channel/UCn0gq8 ... g_8K1nfInQ

light washes out text AND noise in photoshop not in ST

Scan Tailor specific announcements, releases, workflows, tips, etc. NO FEATURE REQUESTS IN THIS FORUM, please.
o3h1p
Posts: 71
Joined: 08 Nov 2010, 22:47

Re: light washes out text AND noise in photoshop not in ST

Post by o3h1p » 29 Feb 2012, 19:52

I think you're right Tulon. The question is how should I deal with this postprocessing because it appears that the resulting pdfs show the noise.

Anonymous2
Posts: 97
Joined: 18 Oct 2011, 16:05

Re: light washes out text AND noise in photoshop not in ST

Post by Anonymous2 » 29 Feb 2012, 22:30

What are you using Scan Tailor for aside from cropping and rotating the image? I used Scan Tailor's "Mixed" mode with the automatic dewarping and got good results with very minimal tweaking. The page looks a little wavy because the book's page isn't being completely flattened by your glass sheet, so this is sort of what it should look like if you were to look at it tangent to the glass plane.

Tell me if you see noise in this picture.
Attachments
IMG_7059.png
Scan Tailor Output

o3h1p
Posts: 71
Joined: 08 Nov 2010, 22:47

Re: light washes out text AND noise in photoshop not in ST

Post by o3h1p » 29 Feb 2012, 22:44

Thanks again for your help Anonymous.

Yes the 'mixed' mode does produce great overall results but it does have some limitations:

* Sometimes does not correctly identify picture zones, it is particularly bad when the diagram/picture has light colors. You can correct this, but this gets to be very time consuming when your book is a thousand pages.
* It also fails to keep colored text colored.

I am amazed at what is is capable of, but unfortunately the 10% failures (just a guess) across a big book with many pictures and complicated diagrams means a lot of corrections, and minimally reviewing each page. On top of that, there is no easy box tool, so you have to click four times each time you want to correct a picture to tell it to be a 'picture zone'. Once you take into account the extra processing time between pages and the above it's not clear if that solution is on the whole better than just messing with the levels in photoshop to wipe out the noise (though the text certainly isn't as crisp using that method (though it does retain text colors)).

Probably my best bet is to try messing with camera settings to get the least amount of noise as possible (higher fstop/shutter speed), with some minor tweaks in lightroom/photoshop if necessary.

Tulon
Posts: 687
Joined: 03 Oct 2009, 06:13
Number of books owned: 0
Location: London, UK
Contact:

Re: light washes out text AND noise in photoshop not in ST

Post by Tulon » 01 Mar 2012, 04:56

o3h1p wrote:The question is how should I deal with this postprocessing because it appears that the resulting pdfs show the noise.
Not surprising, as Acrobat Reader is made by the same company as Photoshop.
Here is a list of things you might try:
* Try opening your PDF in another PDF viewer.
* Try assembling your PDF in another application.
* Try converting your Scan Tailor output files from TIFF to PNG with some application that doesn't do gamma correction. Try IrfanView or Gimp.
* Try to find an option in Adobe Acrobat (or whatever application you use to create your PDFs) to disable gamma correction.
Scan Tailor experimental doesn't output 96 DPI images. It's just what your software shows when DPI information is missing. Usually what you get is input DPI times the resolution enhancement factor.

o3h1p
Posts: 71
Joined: 08 Nov 2010, 22:47

Re: light washes out text AND noise in photoshop not in ST

Post by o3h1p » 01 Mar 2012, 10:20

The conversion to gimp didn't work. I think the only true solution is to remove that noise since I don't really want a pdf that is viewer dependent (i.e. 'make sure you use pdf viewer X, because it will look better'). So disabling gamma in photoshop or acrobat or finding a different viewer isn't really a solution *if* I want the file to be portable. What would be nice is if something (ST, photoshop, irfanview) could take what is displayed in the nice image (with no gamma) and just save that image (like 'save what you see') to use in all viewers, but since that data is really there it seems the choice is really only to hide it or not. There may be some fancy photoshop fu that can be used to remove this, but I haven't found it without washing out other colors.

Thank you both for your help though. I think I really just need to sit down with my scanner and really find out the best settings so I just don't have to worry about this problem.

dpc
Posts: 314
Joined: 01 Apr 2011, 18:05
Number of books owned: 0
Location: Issaquah, WA

Re: light washes out text AND noise in photoshop not in ST

Post by dpc » 01 Mar 2012, 12:40

Is there a copy of the original image (full-res before any processing applied) somewhere that I can download? I have an idea of what's happening but I'd like to have a look at the original raw data first.

Tulon
Posts: 687
Joined: 03 Oct 2009, 06:13
Number of books owned: 0
Location: London, UK
Contact:

Re: light washes out text AND noise in photoshop not in ST

Post by Tulon » 01 Mar 2012, 16:30

The noise is not something added by Scan Tailor. It was present in the original shots and it remains there in Scan Tailor's "Color / Grayscale" output. The real problem is that one or more programs in the following chain "Photoshop -> PDF creator -> PDF viewer" are amplifying it.
o3h1p wrote:The conversion to gimp didn't work.
Please be more specific. Did you load Scan Tailor output directly into Gimp? Did you see amplified noise there?
Scan Tailor experimental doesn't output 96 DPI images. It's just what your software shows when DPI information is missing. Usually what you get is input DPI times the resolution enhancement factor.

o3h1p
Posts: 71
Joined: 08 Nov 2010, 22:47

Re: light washes out text AND noise in photoshop not in ST

Post by o3h1p » 02 Mar 2012, 04:13

@dpc The link is posted above:

http://www.mediafire.com/?92xje8k9ipaz2 ... dphn4jl2sq

@Tulon Sorry I meant "The conversion WITH ifranview didn't work" (when viewed with windows photoviewer). Yes, I agree the noise is in there from the original shot, but if something along the process would save a 'what you see' image (ignore whatever we are calling 'noise'; not that I expect any viewer to since it would be a 'lossy' save, which would be weird) so that the image would look like it does in the ST output screen and wouldn't be amplified in *ANY* viewer, then this would be a solution. As it stands now, the most popular pdf viewer (acroabt) and possibly others (however Anonymous2 created his) will convert these tiffs to the amplified noise if viewed in acrobat. This may be a solution for some ('just use a non-gamma pdf viewer'), but unfortunately it doesn't work for me.

Incidentally, I actually scanned with a flatbed scanner (a single page) and the noise appears there too (the noise being bleed-through from the backside of the page). Here's a portion of the first page (tiff viewed in photoshop):
3-2-2012 2-43-14 AM.jpg
3-2-2012 2-43-14 AM.jpg (18.17 KiB) Viewed 4545 times

* It appears that the post-processing through ST makes the noise *slightly* more visible (probably just particular this particular scan), but the noise is definitely in there in the original. This brings up a few thoughts:


1) If this happens with a flatbed scanner then it appears to be less of a problem with my lighting setup and camera settings (though these might be able to compensate for this problem).
2) I've seen this with a few books, so it seems to be book independent.
3) I'm not sure how many people on here are doing colorful scans (the black-and-whites seem much easier), but I'm surprised no one else ran into this before.

User avatar
Heelgrasper
Posts: 70
Joined: 19 Feb 2012, 21:04
E-book readers owned: None
Number of books owned: 500
Location: Randers, Denmark

Re: light washes out text AND noise in photoshop not in ST

Post by Heelgrasper » 02 Mar 2012, 10:12

I've experienced the problem with getting a shadow of the other side of the page both when making photocopies and flatbedscanning (haven't done much actual book scanning yet). When flatbedscanning I usual have played around a bit with brightness/contrast and managed to tone it down quite good but I no idea how to do that as a batch process.
---
Jakob Øhlenschlæger
Randers, Denmark

The past is a foreign country: they do things differently there
L. P. Hartley

o3h1p
Posts: 71
Joined: 08 Nov 2010, 22:47

Re: light washes out text AND noise in photoshop not in ST

Post by o3h1p » 02 Mar 2012, 12:01

Thanks for the input Heelgrasper.

Photoshop can do this as batch pretty easily. Here are the ways that I've found to get rid of the noise:

* Use the magic wand tool to cut out the noise -- the problem is that this will often cut objects you don't want cut out
* Mess with the levels to clip the high end -- the problem with this is that you wash out the light colors
* Mess with gamma/exposure -- this will often produce a page that is not really white
* Select a region and then perform one of the above -- this works but is VERY time consuming because you have to go through each page

Perhaps the best thing you can do if you're flatbed scanning is to mess with the brighness/contrast on the flatbed so you don't have to worry about photoshop (vuescan has this capability and I'm going to try it now.)

Post Reply