Triplets

tjctf

Task: grayscale PNG where RGB triplets were flattened into single grayscale pixels, with original dimensions hidden in EXIF Comment. Solution: read EXIF for original size, reshape flat grayscale data back into (H, W, 3) RGB array using NumPy.

png numpy image_reconstruction rgb grayscale pixel_flatten exif_comment

exif_metadata_inspectionrgb_channel_flattening_reversalnumpy_reshape

$ ls tags/ techniques/

png numpy image_reconstruction rgb grayscale pixel_flatten exif_comment

exif_metadata_inspectionrgb_channel_flattening_reversalnumpy_reshape

$ cat /etc/rate-limit

Rate limit reached (20 reads/hour per IP). Showing preview only — full content returns at the next hour roll-over.

Triplets — TJCTF 2026

Description

I was messing around with my image and it got really messed up... I see patterns...

Given: chall.png — a 1888×1888 8-bit grayscale PNG (~2.5 MB). The goal is to recover the original image and find the flag.

Analysis

Initial Reconnaissance

Running exiftool on the image reveals key metadata:

Image Width  : 1888
Image Height : 1888
Color Type   : Grayscale
Bit Depth    : 8
Comment      : 2000x594

The image is grayscale (mode 'L'), but the EXIF Comment field contains 2000x594 — the original image dimensions. This is the first critical clue.

Understanding the Encoding

The challenge name "triplets" hints at groups of three — specifically RGB color channels. Each pixel in the original RGB image has 3 values (R, G, B). The encoding process was:

Start with a 2000×594 RGB image → 2000 × 594 × 3 = 3,564,000 individual channel values
Flatten all RGB triplets into a single sequence of grayscale values
Reshape into a square-ish grayscale image: 1888² = 3,564,544 pixels (with 544 pixels of zero-padding at the end)

The description "it got really messed up" and "I see patterns" confirms the image was transformed — the patterns are the interleaved R, G, B values appearing as grayscale stripes.

Solution

The reversal is straightforward:

...