Scan Tailor Adcanced - doesn't recognize text

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

Moderator: peterZ

Post Reply
nightshift
Posts: 24
Joined: 09 Feb 2024, 22:21
E-book readers owned: Nook Glowlight 4, Kindle Fire 5th gen
Number of books owned: 100
Country: USA

Scan Tailor Adcanced - doesn't recognize text

Post by nightshift »

Using Advanced ( this onehttps://github.com/ScanTailor-Advanced/ ... r-advanced) and I'm in the final steps setting output. The book's design is very much like a textbook, where chapter introductions have a pretty picture as a background with text over top. This - plus maybe the colors involved - is causing some problems. (attached images are scaled down)

This is the base page, with most of the scan tailor processing applied:
0006-a.jpg
When processing output with the settings here:
Screenshot from 2024-03-25 13-25-34.png
Screenshot from 2024-03-25 13-25-34.png (32.7 KiB) Viewed 303 times
Plus setting picture areas that leave the text free
I get THIS result
0006-a-bad.jpg
Is there a setting somewhere that I'm missing? Something I should have chosen differently that will give me a foreground split that will include the "Materials" heading so that OCR will see it?
nightshift
Posts: 24
Joined: 09 Feb 2024, 22:21
E-book readers owned: Nook Glowlight 4, Kindle Fire 5th gen
Number of books owned: 100
Country: USA

Re: Scan Tailor Adcanced - doesn't recognize text

Post by nightshift »

Found it! I had to increase the threshold.

The documentation on what each option does, and when you should or shouldn't consider using/changing values, is really light. Most of it is "this feature exists" but doesn't say much about what it does or how to use it.

Anyone have a guide on what everything does?
Post Reply