# AI use cases for Openreality (reality-opened/openreality)

Repo: [reality-opened/openreality](https://github.com/reality-opened/openreality) | Use cases: 5 | Page: https://dedupedai.com/use-cases/reality-opened-openreality/

Open Reality is an open-source toolkit that turns phone video into AI-queryable 3D scenes and exposes them to AI assistants through MCP tools.

This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL.

- Scan a room from phone video and query it: https://dedupedai.com/use-cases/reality-opened-openreality/scan-room-and-ask-assistant/
- Calibrate a scan and measure distances in it: https://dedupedai.com/use-cases/reality-opened-openreality/calibrate-and-measure-scan/
- Plan a walking route through a scanned space: https://dedupedai.com/use-cases/reality-opened-openreality/plan-route-through-scanned-space/
- Export a scan as robot-training data: https://dedupedai.com/use-cases/reality-opened-openreality/export-scan-as-robot-training-data/
- Self-host the scan server on your own GPU or Modal: https://dedupedai.com/use-cases/reality-opened-openreality/self-host-scan-server/

## How to scan a room from phone video and query it with Openreality

Repo: [reality-opened/openreality](https://github.com/reality-opened/openreality) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README

Upload a phone video and ask your AI assistant what objects are in the room and how big it is.

**Who it is for:** A homeowner, renter, or designer who needs quick answers about a physical room without walking around with a tape measure.

**The problem:** Describing a room to an AI assistant from memory or photos gives vague answers about what is there and how large it is.

**The outcome:** A persistent 3D scene of the room that the assistant can query for objects and room size.

### Steps

1. Add the tools to your assistant: `claude mcp add openreality -- npx -y openreality-mcp serve` (or `codex mcp add openreality -- npx -y openreality-mcp serve`).
2. Sign in once with `npx -y openreality-mcp login`.
3. Record a video of the room on your phone, or upload a file such as ~/Videos/kitchen.mp4 by asking your assistant to upload it.
4. Wait for the reconstruction to finish. The README says this takes a few minutes.
5. Ask: "What objects are in my latest scan, and how big is the room?"

### Prompt to paste into your agent

```text
In the openreality repo's MCP tools, upload <video path>, reconstruct it, then list the objects in the resulting scene and report the room size. Verify that the answer says whether dimensions are metric or relative, because the README states numbers are only called metres after calibration.
```

### Caveat

Numbers stay relative until you calibrate the scene with a known real distance, so do not treat room size as metres before that step.

Risk, credentials: The sign-in stores a revocable API key on your machine, so keep that key private.

Risk, reverse engineering: Scans of private rooms are uploaded to the hosted service unless you self-host, so check that you are comfortable with that before uploading.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/reality-opened-openreality/scan-room-and-ask-assistant/

## How to calibrate a scan and measure distances in it with Openreality

Repo: [reality-opened/openreality](https://github.com/reality-opened/openreality) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README

Give one known real distance, then have your assistant measure between points in the scene.

**Who it is for:** A renovation planner, furniture shopper, or contractor who needs a reliable size check before buying or building.

**The problem:** Guessing whether a couch fits against a wall or how far a counter sits from a window is error-prone, and 3D scans without scale give only relative numbers.

**The outcome:** Measured distances and angles in metres, based on a calibration you supplied.

### Steps

1. Add the tools with `claude mcp add openreality -- npx -y openreality-mcp serve` or the Codex equivalent.
2. Sign in once with `npx -y openreality-mcp login`.
3. Pick a real distance you can verify with a tape measure, such as a counter edge to a window.
4. Ask your assistant: "The counter edge to the window is 2.4 m. Calibrate the scene, then measure the couch."
5. Ask for distances or angles between any other points you need checked.

### Prompt to paste into your agent

```text
Using the openreality MCP tools on the scene <scene id>, calibrate with the known distance <distance in metres> between <point A> and <point B>, then measure <target measurement>. Verify that the result is reported in metres only after calibration succeeded, and flag it as relative otherwise.
```

### Caveat

The calibration is only as accurate as the real distance you supply, and the README says server refusals and uncertainty labels are passed to the assistant unedited.

Risk, credentials: The sign-in stores a revocable API key on your machine, so keep that key private.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/reality-opened-openreality/calibrate-and-measure-scan/

## How to plan a walking route through a scanned space with Openreality

Repo: [reality-opened/openreality](https://github.com/reality-opened/openreality) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README

Ask your assistant for a path through the scanned free space from one point to another.

**Who it is for:** A person preparing a space for movement, such as a warehouse coordinator, event planner, or someone checking a route for a mobility aid.

**The problem:** Working out a clear route through a cluttered room or building from photos or memory is slow and easy to get wrong.

**The outcome:** A planned route through the scanned free space, described in words by the assistant.

### Steps

1. Add the tools with `claude mcp add openreality -- npx -y openreality-mcp serve` or the Codex equivalent.
2. Sign in once with `npx -y openreality-mcp login`.
3. Make sure a reconstructed scan of the space exists, either from a new upload or an earlier one.
4. Ask: "Plan a path from the door to the desk and describe it."

### Prompt to paste into your agent

```text
Using the openreality MCP tools on the scene <scene id>, plan a path from <start point> to <target point> through the scanned free space and describe it step by step. Verify that the description names the start and target you asked for and notes any uncertainty the tool reports.
```

### Caveat

The route covers only the free space the scan captured, so areas the video never filmed cannot be planned through.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/reality-opened-openreality/plan-route-through-scanned-space/

## How to export a scan as robot-training data with Openreality

Repo: [reality-opened/openreality](https://github.com/reality-opened/openreality) | Difficulty: Advanced | Time: 1 hour | Basis: From the README

Turn a scanned space into a LeRobot or GR00T dataset, or an Isaac Sim or MuJoCo scene.

**Who it is for:** A robotics engineer or ML researcher who needs a realistic environment and dataset built from a real room.

**The problem:** Building a simulation or training dataset for a real space by hand takes days of modeling and labeling.

**The outcome:** A downloadable zip of robot-training data, saved locally by the MCP process.

### Steps

1. Add the tools with `claude mcp add openreality -- npx -y openreality-mcp serve` or the Codex equivalent.
2. Sign in once with `npx -y openreality-mcp login`.
3. Confirm the scan you want is reconstructed and, if you need metric dimensions, calibrated first.
4. Ask: "Export this scan as robot-training data and save the zip locally."

### Prompt to paste into your agent

```text
Using the openreality MCP tools, export the scene <scene id> as robot-training data in the <format: LeRobot, GR00T, Isaac Sim, or MuJoCo> format and save the zip to <local path>. Verify that the zip exists locally and that the export format matches your request.
```

### Caveat

The README lists robot-training exports as a hosted service, so this use case depends on the hosted service rather than a self-hosted server.

Risk, credentials: The sign-in stores a revocable API key on your machine, so keep that key private.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/reality-opened-openreality/export-scan-as-robot-training-data/

## How to self-host the scan server on your own GPU or Modal with Openreality

Repo: [reality-opened/openreality](https://github.com/reality-opened/openreality) | Difficulty: Advanced | Time: 2 hours | Basis: From the README

Run the full Open Reality server on your own GPU box or Modal account, with no account needed.

**Who it is for:** A developer or team that wants to keep scans of private spaces on their own infrastructure.

**The problem:** Sending scans of private spaces to a third-party hosted service may not be acceptable for some projects.

**The outcome:** A running server you control, reachable by your MCP client through OPENREALITY_URL and a token.

### Steps

1. Clone the repo: `git clone https://github.com/reality-opened/openreality`.
2. Enter the server directory: `cd openreality/server`.
3. Start on your own GPU box with `python -m server.selfhost --data-dir ~/openreality-data`, or deploy to your own Modal account with `modal deploy modal_selfhost.py`.
4. Copy the single token printed at first start, since it is your login.
5. Point the MCP client at your server using OPENREALITY_URL and that token.

### Prompt to paste into your agent

```text
In the openreality repo, follow server/docs/self-hosting.md to start the server at <path or Modal target> and connect the MCP client using OPENREALITY_URL=<your server URL>. Verify that the MCP client can reach the server and that the token printed at first start is not committed to any file in the repo.
```

### Caveat

The self-host paths are described as young, and the reconstruction model the server downloads is licensed for non-commercial use only, so commercial use needs a separate license from its owners.

Risk, credentials: The token printed at first start is your login to the server, so store it securely and never paste it into a shared chat or repo.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/reality-opened-openreality/self-host-scan-server/