Calibr
Log inStart free

Camera calibration & undistortion, fully online.

Run the whole calibration workflow in your browser — detect, solve, undistort, and export.

Start calibrating — free

One calibration workflow

  1. Detect
  2. Validate
  3. Calibrate
  4. Export

No switching tools, no writing scripts, and no losing track of what produced the result.

  1. 01Create project
  2. 02Upload
  3. 03Choose pattern
  4. 04Detect
  1. 05Validate
  2. 06Calibrate
  3. 07Undistort
  4. 08Export
Tune & re-run

See exactly where error concentrates.

The reprojection error map plots every detected corner across the image, shaded by residual magnitude — instantly exposing edge distortion, thin coverage, and the frames inflating your RMS.

Residuals
Per corner
RMS
Per frame
Coverage
Whole sensor

Take results outside CalibrX.

Download JSON or YAML exports, then use the Python SDK to reproduce the same undistortion locally — no manual parameter wiring.

Formats
JSON · YAML
SDK
Python 3.9+
Arrays
OpenCV · NumPy
opencv_pipeline.pypython
from calibrx import load_calibration, undistort

calibration = load_calibration("rectified_calibration.json")
result = undistort(image, calibration)
$pip install calibrx

Solutions

Before vs. with CalibrX.

The same calibration task — done the old way, then the right way.

Before — Manual OpenCV

Hours of guesswork.

  • Manually guess board dimensions, marker sizes, and pattern settings before every run.
  • Trust detections blindly or write custom scripts to inspect weak frames.
  • Repeat separate experiments just to compare Pinhole, Fisheye, and wide-angle models.
  • Debug high RMS values without clear error maps or per-frame quality insight.
  • Lose track of which images, settings, and decisions created the final calibration.
  • Hand-roll fragile export files and OpenCV code for each downstream tool.
With CalibrX

Minutes to sub-pixel.

  • Auto-detect board and pattern parameters for ArUco, ChArUco, chessboard, and circles-grid captures.
  • Validate detections visually, then approve clean frames or remove weak ones before calibration.
  • Compare Pinhole, Pinhole wide, and Fisheye models by RMS error and undistortion quality.
  • Advanced analytics reveal error maps, weak frames, and quality trends to improve results.
  • Versioned calibration history keeps every camera, run, and review decision traceable.
  • Structured exports and the Python SDK handle OpenCV parameters and model-specific undistortion calls.

Pricing Plans

Pay once and keep it for life. Start free, then buy the licence that describes how you will use your calibrations.

Free

See it work

$0/ free forever

Get started with:

  • 50 credits to start
  • Standard pinhole calibration
  • Chessboard, ChArUco, ArUco & circles
  • 1,000 images per project

Fisheye and Wide-angle results stay hidden until you buy a licence.

Start free
Most chosen

Research & Development

Internal work

$49/ lifetime

Everything in Free:

  • Fisheye & Wide-angle results unlocked
  • 300 credits every month, for life
  • Pay once — nothing ever renews
  • Analytics, exports & Python SDK

Not for calibrating products you sell.

Buy R&D

Commercial

Products for sale

$149/ lifetime

Everything in R&D:

  • 1,000 credits every month, for life
  • Pay once — nothing ever renews
  • Usable across your team
  • Every camera model and pattern type
  • Analytics, exports & Python SDK

Full rights, usable across your team.

Buy Commercial

How credits work?

Credits protect expensive compute while keeping calibration costs predictable. Your licence refills them every month, and only actions that create backend work consume credits.

5credits
per project

Create project

Reserves compute and storage for one calibration workspace.

Free
always free

Upload images

Uploads never cost credits, with the same allowance on every licence.

1credit
per detection

Detect pattern

Runs the selected detector and prepares frames for validation.

5credits
per model

Calibrate model

Charged for each camera model solved — Pinhole, Wide, or Fisheye.

Example

One project + detection + three camera models = 21 credits(5 project · 1 detection · 5 × 3 models)

Quick answer

What does CalibrX do?

CalibrX is an online camera calibration workflow for computer-vision teams. Upload checkerboard, ChArUco, ArUco, or circle-grid captures, then let CalibrX detect the calibration target, validate each frame, solve pinhole, wide-angle, or fisheye camera models, and compare the results with RMS error maps and undistortion previews. The platform keeps calibration runs traceable by camera and version, so engineers can remove weak frames, rerun models, and understand why a result changed. Final exports include OpenCV-compatible JSON or YAML with the camera matrix, distortion coefficients, image size, model type, and quality metadata. The workflow is browser-based, but the outputs are built for production pipelines in robotics, machine vision, AR, inspection, and 3D reconstruction. Teams use CalibrX when they need repeatable intrinsics, auditable calibration history, and downstream-ready files without writing OpenCV scripts or buying a MATLAB toolbox. It also helps reviewers repeat decisions across future lens and camera revisions.

Reviewed by the CalibrX engineering team. Last updated June 19, 2026.

FAQ

Common questions.

Everything teams ask before their first calibration run.

How does calibration work in the browser?

Upload your calibration captures, pick a pattern, and CalibrX runs OpenCV detection and the calibration solver in the cloud. You see results, error maps, and exports without installing anything.

Which patterns are supported?

ChArUco, ArUco, Chessboard, and Circles-grid — the standard targets used by OpenCV and most computer-vision pipelines.

Are exports compatible with OpenCV?

Yes. JSON and YAML exports include the camera matrix K, distortion D, image size, and the right distortion model so they drop straight into OpenCV (or any tool that consumes OpenCV conventions).

How many images do I need?

For most cameras, 15–30 well-spread images give sub-pixel RMS. CalibrX shows per-frame quality so you can drop weak frames before solving.

Is my calibration data private?

Yes. Calibration projects live in your workspace, are not shared with other users, and exports stay on your device after download.

Can I calibrate a camera online for free?

Yes. CalibrX runs entirely in your browser — nothing to install — and the free plan covers small calibration projects with no credit card required.

Do I need OpenCV or MATLAB?

No. CalibrX handles pattern detection and the calibration solve for you and exports OpenCV-compatible files, so you never write OpenCV code or buy a MATLAB toolbox license.