I use a program that essentially reads the screen, captures a screenshot in the background, then uses image recognition to run macros through LUA scripts.
I have something like the following image shows: scscreen.png
Currently I have a script that asks the user where the squares are located, places those values into an array and brute force hits each of those screen co-ordinates.
My issue is that each time the game node is activated, the shapes change colour and position on the screen, so it requires constant user input and I cannot simply use static image recognition as its likely to never be the same twice.
I'm wonder if there is a way to do some sort of shape recognition in LUA to detect where on my screen each of those squares are located each time the node is activated to remove the need for the user input so the script can be automated?