What a Free, Open-Source Brilliant Clone Teaches Us About Lazy Eye Training
Someone on Lemmy posted a free, open-source alternative to Brilliant this week: 140+ interactive courses, 1000+ lessons in math, physics, and chemistry, no account, no ads — because the paid version costs up to $160 a year depending on region. It hit 128 upvotes and 70 comments, and the author was upfront that only about a third of pages have been human-reviewed, inviting people to flag what's wrong and fix it in public.
The part worth stealing isn't the price tag. It's the pedagogy: try the thing first, get the explanation after. Interactive practice over passive consumption. That's why it got traction.
I've been thinking about this as an engineering pattern, not just a content pattern. The OpenAha project treats learning like a feedback loop: present a challenge, capture the response, adjust difficulty. That's the same control structure you'd want in any skill-training system — and it's the structure most home vision routines simply don't have.
Where the same gap shows up in vision training
Amblyopia — lazy eye — has a classic home routine: patching. Cover the stronger eye, make the weaker one work, often for an hour a day while a child sits still. It has real clinical history behind it. It also has a real engagement problem, and engagement is what determines whether any routine actually gets done.
Dichoptic training takes a different angle: instead of isolating one eye, you give each eye its own view and force them to cooperate. That's the idea behind AmblyoPunch, a Meta Quest vision-training game.
What you actually do in it
You're in VR. Gabor-pattern coins — oriented, soft-edged striped targets — fly toward you and you punch them with controllers or hand tracking for points. Red spikes come at you; you dodge or duck. You have three hearts, and losing all of them restarts the run. Worlds progress from village to city to space, moon, Mars, Venus, with objects getting faster and smaller while spikes get larger.
If you're a builder reading this, the interesting part is the stimulus design. Gabor patches aren't decoration — they're the classic stimulus in vision science for driving contrast, orientation, and fine-detail processing in the visual cortex. Wrapping them in a punching game is essentially a way to get many trials per session without the user noticing they're doing trials.
It's also a nice systems problem: the game loop and the training loop are two different clocks. The player sees score, combos, and hearts; the training engine underneath sees trial-level accuracy and drives stage transitions. Keeping those decoupled means the difficulty controller can do its job without the player ever staring at a progress bar.
Under the game layer sits a four-stage dichoptic plan, run automatically:
- Monocular warm-up — targets go only to the lazy eye, building tracking and attention without dominant-eye competition.
- Breaking suppression — the dominant eye is heavily dimmed and washed out so the weak eye has to contribute.
- Rebalancing — clarity is gradually restored to the dominant eye across sub-steps while the lazy eye stays active.
- Fusion training — a slight misalignment within Panum's fusion area asks the brain to merge the two views into one 3D percept.
Progression is performance-gated rather than time-gated — typically around 70% accuracy across a rolling 10 outcomes, with demotion possible below 40%, and around 80% required for strabismus profiles before the final fusion stage. That's an adaptive-difficulty system with a hysteresis band, and it's the kind of thing that makes or breaks adherence: too easy and people disengage, too hard and they quit.
Tuned to your eyes
Setup matters. In the wall and wrist menus you choose:
- Lazy eye: Left or Right (picking wrong reverses the intended effect)
- Severity: Mild / Moderate / Severe — changes target size, speed, and stage length; severe profiles get larger, slower targets and longer monocular work, while mild may start closer to binocular training
- Strabismus: Esotropia or Exotropia — shifts Stage 4 disparity and vergence
Automatic mode is the recommended path. Custom mode is there for people who know what they want, with manual monocular/binocular and fusion toggles.
Why the news item matters here
The OpenAha thread is a reminder that people stick with things they can interact with — and that shipping something honest and imperfect in public beats shipping nothing. AmblyoPunch is built on that bet: that daily coordination practice is easier to sustain when it's content someone actually wants to return to, and that real-world confidence (catching a ball, navigating stairs, reading depth) is the point.
Important
AmblyoPunch is a training and assistive game. It is not a medical device, treatment, or cure. It complements — and never replaces — professional eye-care guidance. Consult an eye-care professional for clinical advice about your or your child's vision.
Try it on Meta Quest: https://www.meta.com/en-gb/experiences/amblyopunch/1239507485902689/
Top comments (0)