I was able to implement the MangAI code I mentioned before to improve string chapter detection. This is really useful in separating bonus/special pages from the regular pages, as well as keeping these other pages grouped together.
Next I want to look into improving the page ordering within string and special chapters. For example, it’s not uncommon to see the first page of a string chapter not have a page number, with the second page being denoted with a “2”. I want to try to imply the “1” in the first page to make sure it is properly sorted.
There’s also the issue of handling special pages within special chapters. Currently, MangAI ignores their special page status since it can be hard to know, for example, if “omake” refers to this page being part of a special chapter or if it means a special page within the special chapter. Confused yet? Well, I’ll see what I can do to sort this out in code ^^
Please contact me if you have any suggestions or problems you would like to see improved in MangAI. I pay very close attention to user feedback.
I’ve completed the transfer to a new server. The move was kinda tricky so there may be some bugs. Please contact me if you encounter any problems.
This site will be moving to another server in the next week or so. There may be some downtime during the transition. Sorry in advance for any inconvenience.
Please note that MangAI full license users will not be affected by downtime since full licenses only require activation once per program version. MangAI trial licenses, on the other hand, will not work during downtime as they require constant authentication with the licensing server.
It’s been quiet the past few months since I was busy with a side project.
Currently at work on MangAI MSM improvements. I’m interested in improving MSM string chapter detection. For reference, string chapters don’t use the typical chapter numbering scheme, and instead will be something like chapter “z”. In many cases, this is just another way some scanlators specify omake/extra chapters.
My health has been worse the past few weeks so programming is progressing very slowly. Please email me if you have any feature requests or bugs to report.
New to version 1.14:
-Bug fix: GUI Advanced -> scan_interpretation not working
Small bug, has been fixed.
-Updated helper programs install instructions
ImageMagick now requires installation of Ghostscript (free) if you want to use pdf files as input. Pdf as output is still fine without Ghostscript. Mac and Linux users should already have Ghostscript installed during installation of ImageMagick. Windows users need to download and install it separately.
The way pdftk is distributed has changed. I’ve updated the install instructions.
-New setting: @pdf_input_density
Integer value representing the dots per inch (DPI) of input pdf files.
High density value produces high resolutions when the pages of input pdf files are converted to images for further processing.
High density value, however, greatly increases processing time. Compared to density 72, density 150 takes about 3x longer to process, and 300 takes well over 10x longer!
If text comes out fuzzy, change to a different density and try again.
Acceptable values: 72, 96, 150, 300, 600, 1200
Recommended: 150
Pdf input density used to be fixed at 150. I kept 150 as the default, but you can try lower values to help speed up processing. Depending on the pdf, pages may look poor if you set the density too low compared to the original density they were scanned at.
-AI improvements to MPM and MSM interpretation
Better handling of special volume numbering.
Improved detection of undelineated page numbers (e.g. “c1-3” can be chapter 1 page 3).
Fixed rare crash during MSM scan sorting.
Improved handling when series title contains a number (e.g. “Eyeshield 21”) that shouldn’t be confused with volume, chapter, or page data.
Various logic tweaks.
-Various fixes and improvements
I came across a very rare case where a zip file would not extract properly if a zip entry was wrongfully detected as a file rather than a folder. This is now handled properly.
The program was not allowing trial license requests using emails with uppercase letters in the domain. This has been fixed.
I got a report of .properties settings file being created with some “null” values. I couldn’t reproduce this bug on my test machines, but I modified code that hopefully fixes the problem.
A lot of work went into this update. Most of the time was spent on AI (and subsequent testing of AI). There are a few more AI improvements that I’m working on. They seem to be taking too long and so will have to wait for a future update.
Download the latest version from the MangAI project page.
