From Comment Sections to Dichoptic VR: Building Amblyotube for Kids Who Need a Better View
I first saw the article on a Tuesday morning, coffee cooling on the desk, the glow of the monitor the only light in the room. TechCrunch had written about middle‑schoolers turning NPR’s Spotify comment threads into a secret clubhouse — kids swapping memes, inside jokes, and half‑formed theories in a margin nobody else looked at. Zero onboarding, zero identity checks, just a peripheral UI element they claimed as their own.
It struck me because I’d spent the last year staring at a different margin: the split‑screen of a Meta Quest headset, each eye getting its own texture. I was trying to turn YouTube into a vision‑training tool for amblyopia, the “lazy eye” condition that makes the brain ignore one eye’s input. Traditional therapy — patching, red‑blue glasses, repetitive computer games — feels like homework. Compliance drops after a couple of weeks, especially for teenagers who already have a thousand reasons to look away.
The comment‑section story reminded me that engagement lives where kids already are. They don’t want a new app; they want to keep watching the videos they love, but with a twist that actually helps them. So we asked: what if the “game” is YouTube?
We built Amblyotube on Unity with the Meta Quest SDK, writing a custom dichoptic rendering pipeline that samples the left‑eye and right‑eye textures independently at render time. No post‑processing, no added latency. The shader applies a dominant‑eye shader — essentially digital occlusion — where you can dial blur, contrast, brightness, and opacity. At full opacity it works like a patch, but the sweet spot is partial shading so both eyes stay active.
Then came the lazy‑eye sharpener (MFBF). Using an on‑device AI pass, the app detects human figures in the video and sharpens only the lazy‑eye view. A subtle flicker stimulation rides on those figures — a controlled “jerk” that taps the brain’s motion‑sensitivity pathways. On top of that, a moving magenta circular cue appears only for the lazy eye while a soft grey cue shows to the dominant eye. Magenta was chosen because it rarely shows up in natural footage, making it a reliable fusion anchor.
Recent updates added visual accents: a yellow‑green highlight glow, a red outline silhouette, and pulse controls that breathe at a configurable Hertz rate. The idea is to keep the lazy eye’s attention locked on the target without feeling like a drill.
I remember the first beta tester, a 14‑year‑old named Maya, slipping the headset on in her bedroom. She queued up a cooking channel she watches every night. The first minute she frowned — the magenta cue felt weird, the flicker a little jarring. I watched her adjust the dominant‑eye shader slider, lowering opacity until the stronger eye wasn’t completely blocked. By the twenty‑minute mark she was laughing at the chef’s joke, both eyes tracking the same spoon stir. She told me later, “It didn’t feel like therapy. It felt like I was just watching YouTube, but my eyes finally agreed.”
Safety is baked in. The app is rated 13+, sessions are capped at 30‑40 minutes, never more than an hour, and the onboarding screen forces you to pick which eye is the lazy one — getting that wrong flips the whole effect. The disclaimer is front‑and‑center: this is a training and assistive tool, not a medical treatment, and it’s meant to complement professional care, not replace it.
Seeing Maya’s reaction reminded me of those kids in the comment threads — they didn’t need a new platform, they just needed a space that respected how they already behaved. Amblyotube tries to be that space: a familiar video feed, a few carefully tuned visual levers, and a headset that lets each eye see something slightly different so the brain learns to fuse them again.
If you’re curious — whether you’re a developer interested in dichoptic rendering, a parent looking for a less‑boring way to support vision training, or just someone who likes the idea of turning a comment‑section habit into a health habit — you can try the current build on Meta Quest.
Try it out here: https://www.meta.com/en-gb/experiences/amblyotube/25906906972338493/
Top comments (0)