> For the complete documentation index, see [llms.txt](https://docs.convai.com/api-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.convai.com/api-docs/plugins-and-integrations/unity-plugin-beta-overview/features/vision/usage-examples.md).

# Usage Examples

## Vision in Practice — Implementation Examples

The examples on this page demonstrate end-to-end Vision setups for realistic interactive simulation scenarios. Each example describes the context, lists the components to add, shows the Inspector configuration, and includes any scripting required. All examples assume `ConvaiRoomManager.Connection Type` is set to **Video**.

***

### Example 1: Safety Compliance Training — Live Plant Camera Feed

**Scenario:** A virtual safety officer NPC monitors a live camera feed of a simulated industrial plant floor. Trainees must identify and report hazards; the NPC observes the same camera and provides corrective feedback based on what it sees.

#### Component setup

1. Add `ConvaiVisionPublisher` to the NPC's root GameObject.
2. Add `CameraVisionFrameSource` to the same GameObject.
3. Assign the overhead plant-floor camera to **Target Camera**.
4. Set **Capture Preset** to `HighDetail` (1920 × 1080 @ 30 fps) so the NPC can identify equipment labels and colour-coded safety markers.
5. On `ConvaiVisionPublisher`, set **Publish Policy** to `HighResponsiveness` (15 fps, 1 Mbps) so the NPC reacts promptly to changes on the floor.
6. Leave **Frame Source Component** blank — the publisher auto-discovers the source on the same GameObject.

#### Inspector summary

| Component                 | Field          | Value                  |
| ------------------------- | -------------- | ---------------------- |
| `CameraVisionFrameSource` | Capture Preset | `HighDetail`           |
| `CameraVisionFrameSource` | Target Camera  | *(plant floor camera)* |
| `ConvaiVisionPublisher`   | Publish Policy | `HighResponsiveness`   |

No scripting required for this scenario.

***

### Example 2: Industrial Equipment Onboarding — Webcam Identification

**Scenario:** A technician in a desktop-based onboarding simulation holds up a physical component to their webcam. A Convai NPC identifies the component, confirms its part number, and guides the technician through the installation procedure.

#### Component setup

1. Add `ConvaiVisionPublisher` and `WebcamVisionFrameSource` to the same scene GameObject.
2. Leave **Webcam Device Name** blank to select the default webcam.
3. Set **Requested Width** / **Height** to `1280` / `720` and **Requested Fps** to `15`.
4. Set **Max Output Width** to `1280` and **Max Output Height** to `720`.
5. Set `ConvaiVisionPublisher`'s **Publish Policy** to `AutoCompatible` — 10 fps is sufficient for static or slow-moving objects.

#### Inspector summary

| Component                 | Field              | Value                              |
| ------------------------- | ------------------ | ---------------------------------- |
| `WebcamVisionFrameSource` | Webcam Device Name | *(blank — first available device)* |
| `WebcamVisionFrameSource` | Requested Width    | `1280`                             |
| `WebcamVisionFrameSource` | Requested Height   | `720`                              |
| `WebcamVisionFrameSource` | Requested Fps      | `15`                               |
| `ConvaiVisionPublisher`   | Publish Policy     | `AutoCompatible`                   |

#### Scripting: show available devices in a dropdown UI

```csharp
using UnityEngine;
using UnityEngine.UI;
using Convai.Runtime.Vision.Sources;

public class WebcamSelector : MonoBehaviour
{
    [SerializeField] Dropdown _dropdown;
    [SerializeField] WebcamVisionFrameSource _source;

    void Start()
    {
        string[] devices = WebcamVisionFrameSource.GetAvailableDeviceNames();
        _dropdown.AddOptions(new System.Collections.Generic.List<string>(devices));
        _dropdown.onValueChanged.AddListener(OnDeviceSelected);
    }

    void OnDeviceSelected(int index)
    {
        string[] devices = WebcamVisionFrameSource.GetAvailableDeviceNames();
        _source.SwitchWebcamAsync(devices[index]);
    }
}
```

***

### Example 3: VR Facility Walkthrough on Meta Quest

**Scenario:** A Convai guide NPC accompanies a new employee through a virtual and physical hybrid facility tour on a Meta Quest headset. The NPC observes the real environment through the passthrough feed and points out safety exits, equipment stations, and procedural checkpoints.

#### Component setup

1. Add `ConvaiVisionPublisher` and `QuestVisionFrameSource` to a persistent scene GameObject.
2. Leave **Passthrough Camera Access** blank — the source auto-discovers `PassthroughCameraAccess` in the scene.
3. Leave **Flip Y** enabled (default `true`).
4. Set **Target Frame Rate** to `15` (default).
5. Set `ConvaiVisionPublisher`'s **Publish Policy** to `AutoCompatible`.
6. Ensure the Meta XR SDK is installed and the `PassthroughCameraAccess` component is present and active.

{% hint style="warning" %}
`QuestVisionFrameSource` is only functional on Meta Quest hardware. In the Editor or on other platforms the source enters `Failed` state with `ErrorKind = UnsupportedPlatform`. Use `CameraVisionFrameSource` for Editor testing of this scenario.
{% endhint %}

#### Inspector summary

| Component                | Field             | Value            |
| ------------------------ | ----------------- | ---------------- |
| `QuestVisionFrameSource` | Flip Y            | `true`           |
| `QuestVisionFrameSource` | Target Frame Rate | `15`             |
| `ConvaiVisionPublisher`  | Publish Policy    | `AutoCompatible` |

***

### Example 4: Manual-Trigger Snapshot Session

**Scenario:** A procedural assessment presents trainees with a series of safety inspection tasks. After each task the trainee clicks a **Submit** button, triggering a single burst of Vision publishing so the NPC can evaluate the scene state. Publishing is paused between tasks to reduce bandwidth cost.

#### Component setup

1. Add `ConvaiVisionPublisher` and `CameraVisionFrameSource` as normal.
2. Set `ConvaiVisionPublisher`'s **Publish Policy** to `Manual`. The publisher will not auto-start when the room connects.

#### Inspector summary

| Component                 | Field          | Value      |
| ------------------------- | -------------- | ---------- |
| `CameraVisionFrameSource` | Capture Preset | `Balanced` |
| `ConvaiVisionPublisher`   | Publish Policy | `Manual`   |

#### Scripting: trigger publishing from a UI button

```csharp
using UnityEngine;
using Convai.Modules.Vision;

public class InspectionSubmitButton : MonoBehaviour
{
    [SerializeField] ConvaiVisionPublisher _visionPublisher;
    [SerializeField] float _publishDurationSeconds = 3f;

    public void OnSubmitClicked()
    {
        StartCoroutine(PublishBurst());
    }

    System.Collections.IEnumerator PublishBurst()
    {
        _visionPublisher.EnablePublishing(true);
        yield return new WaitForSeconds(_publishDurationSeconds);
        _visionPublisher.EnablePublishing(false);
    }
}
```

{% hint style="info" %}
With `Manual` policy, `EnablePublishing(true)` uses `AutoCompatible` rates (10 fps, 750 000 bps) unless you call `SetPublishPolicy` to another value before enabling. To use different rates, call `SetPublishPolicy` first:

```csharp
_visionPublisher.SetPublishPolicy(VisionPublishPolicy.HighResponsiveness);
_visionPublisher.EnablePublishing(true);
```

{% endhint %}

***

## Conclusion

These examples cover the four most common Vision configurations: a scene camera feed for simulation monitoring, a physical webcam for object identification, Meta Quest passthrough for mixed-reality guidance, and manual-policy publishing for assessment workflows. Each builds on the same two-component foundation — a frame source and a publisher — with only the source type and publish policy varying by scenario. For scripting APIs, custom frame source implementation, and domain events, continue to [Advanced Topics](/api-docs/plugins-and-integrations/unity-plugin-beta-overview/features/vision/advanced-topics.md).


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://docs.convai.com/api-docs/plugins-and-integrations/unity-plugin-beta-overview/features/vision/usage-examples.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.