What geometric distortions selections to use/choose?

Scan Tailor specific announcements, releases, workflows, tips, etc. NO FEATURE REQUESTS IN THIS FORUM, please.

Moderator: peterZ

Post Reply
Jackson342
Posts: 8
Joined: 11 May 2017, 04:09
Number of books owned: 0
Country: USA

What geometric distortions selections to use/choose?

Post by Jackson342 »

I am unsure of what geometric distortions selections to use for the photo types below:
snapter-book2a.png
snapter-book2a.png (123.78 KiB) Viewed 12278 times

Basically, my output files are coming out like this (I will take off the picture after I have this problem resolved):
sample_output.png

What settings do I need to choose to get rid of the distortions for this image type?

Is there a manual or video that goes through all of the geometric distortions?

Thank you.
Jackson342
Posts: 8
Joined: 11 May 2017, 04:09
Number of books owned: 0
Country: USA

Re: What geometric distortions selections to use/choose?

Post by Jackson342 »

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

Re: What geometric distortions selections to use/choose?

Post by Tulon »

First of all, if you have two pages on a single output image, your "Split Pages" settings are wrong. Dewarping is not going to work on two pages at a time.

The correct distortion mode for your examples is "Curved Lines". However, your case is too hard for automatic dewarping to handle, so you will need to adjust the dewarping grid manually.

As for a video showing how to use dewarping in Scan Tailor, the only one I am aware of is this one. It compares different versions of Scan Tailor when dewarping a complex case. Scan Tailor Experimental is covered starting from 5:30 in the video.
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.
Jackson342
Posts: 8
Joined: 11 May 2017, 04:09
Number of books owned: 0
Country: USA

Re: What geometric distortions selections to use/choose?

Post by Jackson342 »

Tulon wrote: 13 May 2017, 03:31 First of all, if you have two pages on a single output image, your "Split Pages" settings are wrong. Dewarping is not going to work on two pages at a time.

The correct distortion mode for your examples is "Curved Lines". However, your case is too hard for automatic dewarping to handle, so you will need to adjust the dewarping grid manually.

As for a video showing how to use dewarping in Scan Tailor, the only one I am aware of is this one. It compares different versions of Scan Tailor when dewarping a complex case. Scan Tailor Experimental is covered starting from 5:30 in the video.
Thank you so much. I will try your advice. Anyone else have any comments?
Konos93a
Posts: 195
Joined: 19 Sep 2016, 10:00
E-book readers owned: kobo aura,kindle 1,kindle pw3,pocketbook inkpad 2
Number of books owned: 3000
Country: greece

Re: What geometric distortions selections to use/choose?

Post by Konos93a »

if you can have abbyfinereader you could drag n drop there the image and then use commands crop ,split ,deskew and straiten text lines

play with the zoom for better results .the results means more dpi .like common scanners more dpi = better quality = less mistake in ocr.
for more than 300 dpi the ocr is the same

so the dpi are measurable in abby finereader with the command resolution . with the command detect optimal resolution you can see the "real" quality.

cheers
b0bcat
Posts: 49
Joined: 30 Nov 2012, 21:37
Number of books owned: 0
Country: UK

Re: What geometric distortions selections to use/choose?

Post by b0bcat »

@OP - I haven't read your other posts and don't know how you scanned the image e.g. flatbed or camera, but if it were me, bitter experience would reconfirm the rule: time is most efficiently spent in getting the image quality optimal at the image/acquire stage, and not in trying to achieve too much using post-scan software correction. In the case of your sample that might not be so compelling, but there are other more extreme cases of page warp I've seen where just starting afresh makes more sense.

In this case (absent camera, as I have none of the required pixel range) scanning the item would be by using a book edge scanner so you're not scanning two pages at once in landscape or open-book style. Something like the Plustek Opticbook 3600 for instance is pretty cheap to acquire secondhand* and its closeness to the binding margin seems near enough to accommodate this task I would guess, in my experience with that scanner I've got up to about 7mm of the binding margin without causing distortion by tugging on the page though you have to press down firmly to avoid shadow and it can vary according to how thick the book is.

* (on ebay.co.uk for example OpticBook 3600 sales come up often and I've had several near-new for under about gbp35, seems the price is often conditioned by vendors not wanting to post/postage being expensive for large items, and many have been used for one-off projects and discarded by vendors as no longer needed; I can't recommend buying new as the support is in practical terms non-existent so when it fails buy another s/h and/or cannibalise parts as I've done).

The Opticbook 3600 scanner platen is about 12" x 8.5"; if you need bigger the book edge options diminish to near nil secondhand. I've used Microsoft ICE - Image Composite Editor v1.4.4.0 with some success to stitch together scans of part-pages but doing more than one or two pages would be tedious and time consuming - even though I've been amazed at how well ICE works in general.

As for dewarping generally, I've never had much success on occasional attempts so I will be reviewing with interest that video. When I have had minor success it's been in GIMP 2.8 using Tools | Transform Tools | Perspective Tool. In fact I'm using GIMP more and more as I get to grips with it for ad hoc manual fixes of pre- or post- ScanTailor input/output; removing blemishes with paintbrush for instance requires fewer mouse movements than in ST if ST's despecking range isn't enough.
Post Reply