[Concept] Motion detection with autoit

[PragmaThrice]

I think it would be beneficial to detect movement on screen for botting purposes and to allow this to be the basis for pixel based combat scripts (without touching memory).

Premise:
Assume that the character is standing still. Read every ~20th x and ~20th y pixel on screen once, store the results in an array of hex colors and do the same thing again a quarter of a second later. Compare the two arrays of colors to find which (x,y) locations have changed in the last moment.

This wouldn't have to run constantly, only before the bot is about to cast a directional attack. Unless you modified it to only read the outer edges of the screen to detect when new mobs or projectiles enter the area.

The pixel reading response time is the bottle neck. Depending on how fast the pixels can be read (and how many can be read) will determine what this type of feature would be capable of doing.

Thoughts?

[F4L]

The problem with doing this is that there is a fog/mist that moves around on your screen and is constantly changing the color of pixels on your screen, so this wouldn't work. It's makes doing combat scripting in autoit a lot hard without any kind of client side editing or memory scans. You could try doing some kind of imagesearch for monsters/mobs, and tell it what to do based on the type/quantity. You can then use some kind of combat tree, atks closest monster recognized, after screen is clear it loots and repeats etc.