When considering what to share online, a helpful guideline is this: If I wouldn’t want my younger sibling to view it, perhaps it’s best left off the internet. For some, this could apply to a parent or guardian. After hearing my mother thoughtfully assess the writing in the side story from Yakuza 0, I suspect she’d be quite accepting. However, my younger sister would definitely not let me forget any embarrassing moments—and rightly so, particularly if I thought a pixelation filter would keep the existence of my extensive fan fiction hidden.
In a surprising revelation, it appears that decoding pixelated videos is an accessible task. Creator Jeff Geerling—known for his hot dog speaker stunt—recently issued a challenge to his audience to reveal the contents of a network share obscured by pixelation, incentivizing them with a $50 reward for their success. Notably, his viewers took up the challenge and succeeded in three distinct yet equally unsettling methods.
Geerling elucidates that while three individuals approached the task differently, they all relied on a common idea. He explains, “The premise is that pixelation functions similarly to shutters over an image. As you shift the obscured image beneath, you can catch glimpses of different areas. If you maintain a fixed reference point, such as a window of consistent size, you can gather pixel data from the concealed picture.”
With a series of fragmented images, anyone motivated enough and utilizing AI technology can piece together what you aimed to obscure with pixelation. Personally, I equate this method to restoring a kaleidoscope to its original state.
Geerling mentions that if he hadn’t moved the pixelated area in his video, it might have become considerably more challenging for viewers to decipher, though not entirely impossible.
With advancements in neural networks and AI, it has become fairly straightforward for computers to analyze and make sense of apparent disorder, as opposed to the prior need for extensive processing power and extensive knowledge of the subject.
So, if pixelation is inadequate, what alternatives do we have? Geerling suggests that conventional blurring might not guarantee safety either. He plans to obscure sensitive information in future content using a solid color mask, minimizing the visual detail available for neural networks to analyze. This approach could mimic the style of redacted documents commonly seen in suspenseful thrillers. Alternatively, utilizing emojis might be a creative solution, though it would certainly alter the tone of a narrative in games like Control.
