SEETF 2022 Writeup

Only some challenges are available in this writeup, for the purpose of submission.

Welcome

Misc

Welcome

Author: zeyu2001

Description

Welcome to SEETF! Submit the teaser video flag here.

https://www.youtube.com/watch?v=0GVC30jiwJs

Download: https://drive.google.com/file/d/1Dlfp6vC1pto7vW5WUKqvX-tbIt6tfGfg/view

MD5: 8356ab36b1bd286a514dfdca483207e8

While watching the wonderful trailer, we notice that starting from 00:00:26, there are some weird blinking white pixels near the top-right corner.

Observe the top-right corner, there are some blinking pixels in seemingly random positions.

To make running and debugging any processing code faster and easier, we can crop the video to only keep the top-right corner and also remove the front part of the video using FFmpeg:

ffmpeg -ss 00:00:26 -i Trailer_With_Challenge_crop.mp4 -filter:v "crop=180:180:1740:0" processed.mp4

-ss 00:00:26: input seeking to start from 00:00:26
-i Trailer_With_Challenge_crop.mp4: input file
-filter:v "crop=180:180:1740:0": crop the video to only keep the top-right corner, 180:180:1740:0 being w:h:x:y, with w and h being the width and height of the cropped video, and x and y being the x and y coordinates in the original video to be used as the top-left corner of the output video.

We then try to combine all the pixels together, to see if it produces any useful information. To do so, I decided to use OpenCV as I was decently familiar with the library.

import cv2 as cv
import numpy as np

# Create a blank image of the same size as the input video
output = np.zeros((180, 180), np.uint8)

cap = cv.VideoCapture("Trailer_with_challenge_crop_trim.mp4")  # Open the video file
while cap.isOpened():  # Loop until the video is over
    # Read the next frame into frame, ret will be `false` if no frame is read
    ret, frame = cap.read()

    # Break if there are no more frames to read
    if not ret:
        break

    # Filter the white (BGR above (200, 200, 200)) pixels from the frame
    mask = cv.inRange(frame, (200, 200, 200), (255, 255, 255))
    # and put them in the output image, using or to combine the different frames
    output = np.bitwise_or(output, mask)

cv.imwrite("output.png", output)  # Save the output image

cap.release()  # Close the video file (er actually unnecessary but i think good habit)

The script above will generate a output.png file, which contains all white pixels (those with BGR values all above 200) as pure white pixels. This combines all the blinking white pixels, producing a scannable QR code.

Scanning the QR code gives us the flag (SEE{W3lc0m3_t0_SEETF_95c42d3be1cb93cce8241235529ad96f8e0e1c12}). However, the QR code is not very aesthetically pleasing… With the knowledge that the pixles produce a QR code, we can actually improve our code to recover a perfect(?) QR code.

We notice that in every frame, the white pixels actually come as a square of 3x3 pixels. This corresponds to one square in the QR code grid. Therefore, we can resize every frame from 180x180 pixels to 60x60 pixels, taking the average of a 3x3 square of pixels to produce every pixel in the new frame.

This can be done with cv.resize().

import cv2 as cv
import numpy as np

# Create a blank image of the same size as the input video
output = np.zeros((60, 60), np.uint8)

cap = cv.VideoCapture("Trailer_with_challenge_crop_trim.mp4")  # Open the video file
while cap.isOpened():  # Loop until the video is over
    # Read the next frame into frame, ret will be `false` if no frame is read
    ret, frame = cap.read()

    # Break if there are no more frames to read
    if not ret:
        break

    # Resize the frame with `interpolation=cv.INTER_AREA` to get a frame
    # with each pixel representing a square in the QR code grid
    resized = cv.resize(frame, (60, 60), interpolation=cv.INTER_AREA)
    # then filter the white (BGR above (200, 200, 200)) pixels from the resized frame
    mask = cv.inRange(resized, (200, 200, 200), (255, 255, 255))
    # and put them in the output image, using or to combine the different frames
    output = np.bitwise_or(output, mask)

cv.imwrite("output.png", output)  # Save the output image

cap.release()  # Close the video file (er actually unnecessary but i think good habit)

Modifying the script slightly to work with a shrunk frame, we can get a perfect QR code (open image by itself if it’s too small, since it’s only 60x60 pixels).

ta-da