PiFind

This searches the digits of π for a sequence of bytes which can match the pattern in a target image, so that a palette can be chosen to approximate the target image using a contiguous sequence from π as the pixel data.

See https://kundor.github.io/Finding-Waldo/ for an example.

The script requires the Python package Pillow. You can probably install it by running the command

python3 -m pip install Pillow.

Otherwise, I've written it to only use the standard library, so it should run with Python 3 (probably with versions 3.6+).

1. Download the script pifind.py (Note: you only need this file, not the whole repository.)
2. Optionally, download pi_hex_1b.zip from archive.org/details/pi_hex_1b and place it in the same directory. This will speed up the search. Otherwise, digits will be fetched from the Web at https://pi.delivery.
3. In the same directory, place the image to search for as target.png. It should be not much more than 500 pixels (around 22x23 or smaller) and use at most 6 distinct colors.
4. Run python3 pifind.py

The "best" image found so far will be saved as found.gif.

What the script considers "best" might not match what a human thinks looks good. You might want to check out the candidates as they're found and save ones you like under a different name.

The script output includes the digit index where the current best image starts. You probably want to make a note of that.

PiFind has been tested on Ubuntu, Funtoo, and Windows (in cmd.exe), using Python versions from 3.6 to 3.10, so with luck it will work for you.

Some additional little utility scripts are present:

• verify.py verifies if any image file contains hex digits from π starting from a given index.
  Call it like python3 verify.py <index> <image file name>
• makeimage.py creates an image approximating a given image with hex digits starting from a given index. This is handy if you already know the offset and don't want to spend time searching through π.
  Call it like python3 makeimage.py <index> <target image> <new image>
• haystack.py creates a giant image of the first 100 million hexadecimal digits of π, colored with the palette from a found image, and aligned so that the image shows up. The image has to be present in the first 100 million hex digits! This one requires pi_hex_1b.zip to be present, unlike the other scripts (which require an internet connection instead).
  Call it like python3 haystack.py <found image> <new image>
• depalette.py removes the palette from an image and saves the same data interpreted as grayscale and with the Plan9 palette.
  Call it like python3 depalette.py <found image> <new grayscale image> <new plan9 image>