> For the complete documentation index, see [llms.txt](https://docs.theseus.us/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.theseus.us/vozilla-gcs-software/vozilla-app/camera/camera-calibration.md).

# Camera calibration

<figure><img src="https://content.gitbook.com/content/hzvlBbeiwnrjdGDAaBcQ/blobs/tlqQ95onIhr5y70WHMJg/Recording%202026-08-26%20122519.gif" alt="" width="542"><figcaption></figcaption></figure>

The Camera Calibration tool in Vozilla lets you generate and save a calibration file (camera intrinsic) for your camera. A good calibration improves the accuracy of Cyclops tracking and reduces distortion, especially near the edges of the image.

#### What you need

* A 7×8 asymmetric circle grid calibration target, found below.

{% file src="/files/9jIZU1cdlLUHxN8zqzdi" %}

* For thermal cameras, we have found it works best to 3D print the grid and place it on a uniform warm surface so the target is clearly visible. A .STEP file for this grid can be found below.

{% file src="/files/YUxFEpZH5geflfCvGavu" %}

{% hint style="info" %}
Placing the 3D print against a monitor works well for thermal cameras, providing a consistent warm background.

<img src="https://content.gitbook.com/content/hzvlBbeiwnrjdGDAaBcQ/blobs/Xu3D5UChlUD2wFcw5p5W/image.png" alt="" data-size="original">
{% endhint %}

#### Capture calibration images

1. Connect your camera and open the Camera Calibration page.
2. Use Capture Frame to record images of the detected grid.
3. Move the target through a variety of:
   * angles (tilt and rotate)
   * positions (across the full frame, including corners)
   * distances (close, medium, far)

For best results, capture 20+ frames in which the pattern is cleanly detected.

<figure><img src="https://content.gitbook.com/content/hzvlBbeiwnrjdGDAaBcQ/blobs/TfiAj4cptXhkrLSYx2Ug/Untitled%20design(1).gif" alt=""><figcaption></figcaption></figure>

{% hint style="warning" %}
If your target is white dots on a black background, enable **Inverted Pattern** so detection works correctly.
{% endhint %}

#### Compute and save

After capturing enough frames, Vozilla will prompt you to compute the calibration. Save it using the default name or a descriptive name (helpful if you switch cameras or lenses later).

#### Verify quality

Use the Distortion toggle (bottom-right of the camera view) to compare the calibrated image to the raw image. A good calibration should reduce warping and keep lines and features looking consistent across the frame.

<figure><img src="https://content.gitbook.com/content/hzvlBbeiwnrjdGDAaBcQ/blobs/reLIIQiqA6DC3AHh0aOE/Untitled%20design(3).gif" alt=""><figcaption></figcaption></figure>