He finally confronts the love interest. As she speaks, the screen splits: left side is her actual face (uncompressed, raw, messy), right side is his internal "decoded" version—smooth, idealized, lacking pores or tears. When she says, "You don’t even see me," the right side glitches violently into a gray block of corrupted data. The codec crashes. For three seconds, the screen goes black. No audio. No motion vectors. No compression.
This leads to his first major mistake: because he only tracks changes, he fails to notice a crucial detail—her meeting with an old friend. The codec drops that macroblock as "unchanged background," and he misinterprets a platonic hug as a romantic betrayal.
Picking up immediately after the premiere’s reveal, Episode 2, "OpenH264," deconstructs the series’ central metaphor: the act of watching someone is never lossless. The episode’s title references the open-source video codec widely used in WebRTC, Zoom, and browser-based recording—a tool that compresses raw visual data into a streamable, viewable format, but at the cost of dropping subtle frames, introducing blocky artifacts, and smoothing over critical detail.
The episode opens with a close-up of a security camera’s lens, its red recording light flickering. Our protagonist is reviewing raw footage from a coffee shop’s NVR (Network Video Recorder). He freezes on a single perfect frame of the love interest—what codec engineers call an I‑frame: a complete, uncompressed image that all subsequent predictions will rely on. "This," he whispers, "is the only honest second. Everything after this is just... difference data."
As he follows the love interest through her day, the screen visually distorts. Motion vectors appear as faint cyan lines trailing her movements. The audio occasionally glitches—a word repeated, a laugh truncated. The narrator explains: "A P‑frame doesn’t store the whole picture. It just stores what changed since the last frame. That’s how I see her now. Not whole. Just the difference between what I want and what I saw."
When the image returns, it’s in baseline profile—no B‑frames, no predictive frames, just a single frozen I‑frame of his own reflection in a dark window.
A monologue occurs while he watches a video call she has with her mother, recorded via WebRTC (which often uses OpenH264). The video drops to 240p, stutters, and loses audio sync. He says: "They tell you high bitrate means high truth. But bandwidth is a lie. Even at 4K, you’re only seeing 24 still images per second stitched together by a lie called persistence of vision. Love is the same. We fill the gaps between frames with assumption. I assumed she was happy. I assumed she wanted me. But those were just... interpolated frames. Guesswork." Climax – The Lost I‑Frame
In this episode, our narrator (You) is no longer just a passive observer. He has begun "encoding" the people around him—forcing complex, messy human beings into a low-bitrate, H.264-compliant version of themselves that fits his own narrative. The episode asks: When you compress a person into an object of obsession, what gets lost in translation?