Building a Simple Python Bot for a 4399 Mini‑Game Using autopy and PIL

The article introduces a small Python project that automates a 4399 sushi‑making mini‑game by simulating mouse clicks and analyzing on‑screen images. It starts by recommending the installation of autopy , PIL , and pywin32 on Windows.

It explains how to capture screenshots using either autopy or PIL.ImageGrab , and how to locate fixed UI elements (customer avatars, ingredient slots) by measuring screen coordinates. The author notes that screen coordinates start at (0,0) in the top‑left corner and discusses the inaccuracy of autopy 's absolute coordinate conversion, showing the original and corrected formulas:

point.x *= 0xFFFF / GetSystemMetrics(SM_CXSCREEN);

point.x = point.x * 0xFFFF / GetSystemMetrics(SM_CXSCREEN);

Mouse movement is performed with autopy.mouse.move(x, y) , and a short sleep is recommended to let the game react before issuing the next click.

For image recognition, the tutorial moves beyond the exact‑match find_bitmap method (which fails with minor pixel differences) and adopts a perceptual‑hash approach. It resizes captured avatars to a fixed size (e.g., 18×13), computes a binary fingerprint, and compares fingerprints using Hamming distance:

def hamming_dist(self, hash1, hash2): return sum(itertools.imap(operator.ne, hash1, hash2))

If the distance exceeds a threshold (e.g., 50), the script assumes no dish is present at that position. Otherwise, the smallest distance identifies the requested dish.

Finally, the bot stores a mapping of dishes to ingredient coordinates and programmatically clicks the required ingredients in order, encapsulated in a simple Python class (code omitted for brevity). The author emphasizes that the bot is a learning exercise rather than a competitive cheat.