> For the complete documentation index, see [llms.txt](https://docs.unitlab.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.unitlab.ai/solutions/multimodal-patterns-guides/video-documents-and-3d-context.md).

# Video, documents, and 3D context

Label video next to the manual, sensor log or 3D scan that explains it, so every decision has its context.

A video often needs a second source to be labeled correctly: the work instruction that defines the steps, the sensor log that confirms an event, or the 3D scan that shows depth. Unitlab opens them together.

<figure><img src="https://2337177515-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fi6jXAxtoN6UXTLAJFexq%2Fuploads%2FNbU8y6pqUzG6nhT7ruh3%2Fsol-mfg-smartwatch-station.webp?alt=media" alt="One test episode: head-camera video, start frame, test-pad current plot and work-instruction PDF in a 2 x 2 layout" width="800"><figcaption><p>Video, image, sensor plot and PDF of one test episode in a single task.</p></figcaption></figure>

## Common pairings

| Primary data      | Context                | What the context answers                 |
| ----------------- | ---------------------- | ---------------------------------------- |
| Operator video    | Work-instruction PDF   | Did the operator follow step 3?          |
| Operator video    | Sensor CSV             | When exactly did the part touch the pad? |
| LiDAR point cloud | Six calibrated cameras | Is this cluster a car or a hedge?        |

<figure><img src="https://2337177515-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fi6jXAxtoN6UXTLAJFexq%2Fuploads%2FoT4Pmfumuf7eJeQpbRD4%2Fsol-reuse-lidar-hero.webp?alt=media" alt="A LiDAR frame with a car cuboid and six camera images showing the projected cuboid" width="800"><figcaption><p>In LiDAR scenes the cameras are part of the scene itself: each cuboid is projected into every image.</p></figcaption></figure>

## Set it up

1. For files of one event, group them into a Data Group with a 2 x 2 layout. See [Multi-camera and multi-angle work](https://docs.unitlab.ai/solutions/multimodal-patterns-guides/multi-camera-and-multi-angle-work.md).
2. For LiDAR, upload the scene with its calibration; the cameras come with it. See [Autonomous systems](https://docs.unitlab.ai/solutions/industry-solutions-guides/autonomous-systems.md).
3. Label in the primary tile and use the others to confirm. Mark sensor peaks as intervals so they can be matched to video frames.
