Lensboy – Lightweight camera calibration with spline distortion models
Spline distortion beats OpenCV for cheap lenses; swappable API outclasses mrcal's overhead.

Rotation-invariant calibration that stays zero when you rotate the phone — most apps fail this.
iPhone users doing DIY projects, home improvement, or furniture alignment
Apple Measure app · Bubble Level: Pro Measure Tool
I built a simple, single-purpose iOS utility: a bubble level that lets you calibrate out the physical angle offset caused by iPhone camera bumps and cases. It automatically switch between 3 modes when you try to use it on the back, along the long edge, or along the short edge.
The Problem with the iPhone's Measure app and other Bubble Level apps:
If you lay a modern iPhone flat on its back to see if a surface is level, the camera bump (and the shape of the phone case) tilts the device by a degree or two. Existing apps do not offer a correct way to calibrate this physical offset away, and are often annoys you with Ads, paywall, etc. Their so called calibration allow you to click a button to set the reading to zero, but when you rotate your phone on the same flat surface, it won't be zero anymore. The correct calibrated reading should be zero no matter how you place the phone.
The Solution:
I wanted a clean tool that solves this specific issue. To use it, you place your phone on a known flat surface, tap "Zero" and follow the instruction to simply rotate your phone a few times. Now you can use your iPhone as a leveler as if it has a flat case.
A Few Details:
- Free: The calibration feature is free to use, a one time payment of $1.99 to enable sub-digit reading and haptic/sound feedback. - Privacy: The app collects zero user data and requires no sign-ups or network permissions. - UI: It is designed to be minimal and lightweight, focusing only on the utility of leveling.
The app is free to download: https://apps.apple.com/us/app/id6767735397
I would appreciate your feedback on the usability, the calibration flow, or any additional utility features you think would be useful to add.
Thanks!
Spline distortion beats OpenCV for cheap lenses; swappable API outclasses mrcal's overhead.
Trivia game measuring confidence calibration, but calibration training already exists (Metaculus, Good Judgment).
Lightweight Rust TUI that works with existing Calibre libraries; instant search beats Calibre's sluggishness.
Calibration game feeds behavioral dataset on human perception blind spots.
Bubble sort on a Turing Machine with 31 states; educational but purely academic novelty.
Predict-calibrate extraction reduces noise, but Zep and Mem0 already dominate the agent memory space.