Skip to content

Latest commit

 

History

History
104 lines (78 loc) · 3.34 KB

colours.md

File metadata and controls

104 lines (78 loc) · 3.34 KB
Raw

Somebody leAked TrumP's favorite colors, looks like they used a really esoteric format. Some chiNese hacker named "DanGer Mouse" provided us the leak, getting this crucial info could really sway voters at the polls!

The following image was provided for this challenge:

colours

For this challenge, there were quite a few clues in the description.

  • The odd capitalisation of certain letters was the first - the letters APNG were capitalised when they didn't need to be. This hinted that the image was in fact an APNG - opening it in Firefox showed the animated nature of the image - confirming that it was an APNG.
  • "esoteric format" hinted that perhaps an esolang was involved.
  • "Danger Mouse" is the author of the esolang Piet.

After realising the image was a Piet program, I downloaded npiet and executed the program, the output was:

#ff0000

I realised that each frame in the APNG was likely to be a different Piet program. With this in mind, I tried some online APNG splitters, but eventually settled on using GIMP.

I installed an APNG plugin and opened the image. After waiting for all 16384 frames to be converted into layers, I looked for a way of exporting each layer as a separate PNG. I eventually settled on Export Layers. Unfortunately, exporting as a PNG wasn't working so I exported each layer as a TGA file and converted to a PNG later.

After exporting all of the layers as individual images, I tried to use npiet to execute one. This didn't work, as the layers were just deltas, rather than complete images. To solve this, I edited the original PNG into a base image and composed the layers on top of it using ImageMagick (I also converted the background of the layers from white to black, although I am not sure if this was necessary).

Base

I wrote the following script to convert, compose and execute each layer:

#!/bin/bash
for f in *.tga
do
    fpng="$(basename "$f" .tga).png"
    echo -n "$fpng : "
    convert -fuzz 5% "${f}" -fill black -opaque white "${f}"
    convert base.png "${f}" -compose lighten -composite "${fpng}"
    npiet "${fpng}"
    echo
done

I executed the script in the following manner:

./process.sh | sort -V | sed 's/^[^#]*#//' > out.txt

I then had a file that contained hex colour codes. The next step was to write a Python script to convert the text file into a complete image.

#!/usr/bin/env python3
import struct
from PIL import Image

output = Image.new('RGB', (128, 128))
colours = []

with open('out.txt', 'r') as f:
    for colour in f.readlines():
        colours.append(struct.unpack('BBB', bytes.fromhex(colour.strip())))

output.putdata(colours)
output.save('out.png', 'PNG')

The generated image looked promising, but running npiet on it didn't yield a flag.

out

On closer inspection, the darker coloured blocks linking up the columns were missing. The curve on the right hand side of the image was also incomplete. After fixing the issues, I was left with the following image:

out-final

Running npiet on this image and entering a random string resulted in the following output:

fag{7h15_w45n7_3v3n_4_ch4ll3n63._54d_.}

After adding the missing "l" into the word "flag", putting the flag into the site yielded 250 points for the team.