I’ve hit this exact weird low-level error before when working with Pillow and older Python versions—let’s walk through the most likely fixes for your code:

1. Your x/y coordinates are probably non-integers #

Pillow’s putpixel expects the coordinate tuple to be made of whole integers. If cx.row or cx.col contain float values (even something like 2.0 that’s technically an integer), passing them directly will trigger this SystemError from Pillow’s C extension backend—since it’s expecting strict integer types for coordinates.

Quick fix: Convert x and y to integers explicitly:

im.putpixel((int(x), int(y)), v)

2. Double-check the type of your color value #

Even though you’re using an 'L' (grayscale) image, if cx.data returns tuples instead of single integers (like (128,) instead of 128), or if color is a numpy numeric type (like numpy.int64), subtracting it from grayscale_color might create a value that Pillow’s C API doesn’t recognize properly.

Fix: Cast color to a native Python integer before calculating v:

v = grayscale_color - int(color)

3. Verify Pillow version compatibility with Python 3.6 #

Python 3.6 is end-of-life, and newer Pillow versions (9.0+) dropped support for it. If you "updated" Pillow to a version that doesn’t play nice with Python 3.6, that could cause odd low-level errors like this. Try rolling back to the last Pillow version that supports Python 3.6:

pip install pillow==8.4.0

Debugging step to confirm the issue #

Add quick print statements inside your loop to inspect the types of your variables—this will help you pinpoint exactly what’s causing the problem:

for x,y,color in zip(cx.row, cx.col, cx.data):
    print(f"x type: {type(x)}, y type: {type(y)}, color type: {type(color)}")
    v = grayscale_color - int(color)
    print(f"v type: {type(v)}, v value: {v}")
    im.putpixel((int(x), int(y)), v)

内容的提问来源于stack exchange，提问作者user80167