Jump to content

Recommended Posts


Hmm, not going to make so many lines to Introduce it all.

I have made a new toy that can be used by everyone who needs to add some Text Recognition parts to this BOT.

This toy is at least 2X faster than the Official toy (i mean the Official Developers only limited one) according to the benchmark comparison tests we got.

This is a simple, small and lightweight .dll file that can be used in AutoIt by DllCall.

The .dll itself is not open-sourced yet! i mean it might be, who knows...?!


  • How-To

An example .au3 file has been included to the Archive to let you know what parameters it takes and how to use it in AutoIt.



  • How-To make Bundles/OCR files


  1. Crop the Char you want to recognize from an Image
  2. Erase (Make transparent) the pixels of the cropped image that usually has different colors in another scenes and images
  3. Your cropped image file will be something like the below, save that as a 32-bit png file.


The ocr file name has a template! that is:


Char: The value that will be returned if it recognize.

MaxDif: The Maximum Difference that the Red, Green, Blue or Alpha that is allowed to have in comparison to the Source image, If Source image's first pixel is (Red: 200, Green: 215, Blue: 217) and the OCR image's first pixel has (Red: 180, Green: 200, Blue: 200), with a MaxDiff of 20, it will be a match! Note: Transparent pixels are matches anyway.

SomeNumber: Used to make it possible to have multiple files with same char and maxdiff. NOTE: I didn't made it like it delete Duplicated matches in same positions, But why?: That way you're forced to make your bundle carefully and not with bunch of useless ocr files for the same char, results less Ocr files which means a faster recognition


It can also read a long width image and delimit the OCR values with " | "


I have also made a GUI application for the library to simplify testing bunch of Source images to an OCR Bundle, Creating Bundle files and Extracting them.



Everyone can use this toys to extend the OCR functionality of the BOT or their MOD.


Thanks to the AIO-MOD team for testing this, they have also implemented this all in their MOD so you can use their Github repo. to see how to implement it in your own MOD or Compilation.


  • Download




Let me know if you have found issues or bugs right in here.

Edited by LookMomImOnTheWeb
v1.0.6.0 - GUI
  • Like 1
  • Upvote 3
Link to post
Share on other sites
Posted (edited)

Updated the first post with GUI v.

- Fixed the "Extract Bundle" button.



Updated the first post with GUI v.

+ Added 'Status' Column to the Grid view to make it possible to sort by Status (NotChecked, IsMatch, NotMatch)

Edited by LookMomImOnTheWeb
Link to post
Share on other sites

Updated the first post with Dissociable.OCR.dll v.

- Added TrainImage function.


Dissociable.OCR.GUI v.
- Added GUI options for TrainImage function.



Q: What's TrainImage function? huh?

A: A function to Train OCR image files!


Q: Woah! How you made such a function that needs so much brain and focus?

A: Well, No AI or things like that. FOCUS! The function goes through each template (all png files in directory entered) within specified rectangle and ONLY keep the pixels that matched with the MaxDiff you specified in ALL templates. Imagine you collected 200 images of '2' in specific rect of COC. you make a Source image of that '2' (Simple crop), then set RECT and templates directory and it will make you a .png file that was able to recognize the '2' in all those 200 template images, whether the background changed or not! it will get you a template that match 'em all. which MIGHT result in accurate bundles, not guaranteed... the resulted images always has to be tested.

The resulted image might even be a full transparent png file, but guess what! that matches all! genius! JK, delete 'em.


Q: More details?

A: Hover on the fields, count to 1, Tooltip is with you.

Edited by LookMomImOnTheWeb
Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

  • Create New...