> For the complete documentation index, see [llms.txt](https://multiset.gitbook.io/multiset/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://multiset.gitbook.io/multiset/basics/georeferencing-maps/auto-georeference-map.md). # Auto Geo-reference a Map The MultiSet App can geo-reference an outdoor map for you, with no Unity work and no manual satellite alignment. You walk the mapped area, localize at several well-spread points, and the app pairs each successful VPS localization (the camera's pose in the map's local frame) with the device GPS fix at that instant. When you tap **Done**, all of these point pairs are submitted as control points and the server solves the map's geo-reference: its origin in WGS84 (latitude, longitude, altitude) and its heading relative to true north. This is the recommended workflow for **outdoor maps captured in areas with a clear sky view and good GPS reception**. For indoor maps, or when GPS is unreliable, use the manual [How to Align Scans](/multiset/basics/georeferencing-maps/how-to-align-scans.md) method instead. *** ## How it works Each captured point is a single correspondence pair: * **Local pose** : where the camera is inside the map, obtained by running a VPS localization at that spot. * **GPS fix** : the device latitude, longitude and altitude, snapshotted at the exact moment the localization frame is taken. The GPS fix is captured at the instant you press the capture button, so the two halves always describe the same physical spot even if you shift slightly while waiting. With enough of these pairs spread across the map, the server can compute the single rigid transform (position and heading) that best maps the local coordinate frame onto real-world coordinates. More points, better spread, and stronger GPS all improve the fit. *** ## Prerequisites * The outdoor map is already scanned, processed, and shows as **Active** in the MultiSet App. * You are physically at the mapped location with a **clear view of the sky** (open areas, away from tall walls, dense tree cover, or covered parking). * **Location services and precise location are enabled** for the MultiSet App. * You allow a short time (typically 15 to 60 seconds after opening the screen) for the GPS to settle to a good, stable accuracy. {% hint style="warning" %} Geo-referencing accuracy is limited by your GPS accuracy. Capturing points while GPS is still poor (large accuracy radius) produces a weak or rejected solve. Always wait for a good fix before capturing. {% endhint %} *** ## What makes a good capture point The quality of the final geo-reference depends almost entirely on **where** you capture points and **how good the GPS is** at each one. Keep these five rules in mind. They are also shown on the in-app instructions screen. 1. **Stand still while capturing each point.** Movement while the fix and localization are taken adds error to the pair. 2. **Wait for a good GPS signal before capturing.** Check the GPS status before each point (see below). 3. **Spread points across the entire mapped area.** Do not bunch them in one corner. Aim to cover the map in every direction from its centre. 4. **Face a different direction at each point.** Varying the viewpoint gives the solver more independent information. 5. **Maximise the distance between points.** The farther apart your points are, the more accurately the heading (rotation to true north) can be resolved. Two points a metre apart barely constrain heading; points at opposite ends of the map constrain it well.

The in-app guidance: spread well-separated points across the whole map, not clustered in one spot.

{% hint style="info" %} **Why spread and distance matter.** Think of it like a survey. A few widely separated, well-measured points pin the map firmly in the real world. Many points crowded together, or all facing the same way, leave the position and heading loosely defined and the solve error grows. {% endhint %} *** ## Step 1: Open the Geo-reference tool 1. Open the **MultiSet App** and sign in. 2. Find your map to geo-reference. 3. Tap the map's options menu (the three-dot icon), then select **Geo-reference**.

Open the map's menu and choose Geo-reference.

*** ## Step 2: Read the instructions and start The **Geo Referencing Map** screen opens with a **Keep In Mind** panel summarising the five capture rules and a diagram of well-spaced points across a map. Read it, then tap **Next** to open the capture camera. *** ## Step 3: Wait for a good GPS fix At the top-right of the capture screen is the **GPS** status button. Before capturing any point, make sure the fix has settled: * Stand still in an open spot and give the GPS a few seconds to stabilise. * Tap the **GPS** button to check the current fix. Wait until the reported horizontal accuracy is good (roughly **±5 m or better** is excellent; **±10 m** is acceptable) and the fix is fresh, not frozen. * If accuracy stays poor, move away from buildings, walls, or tree cover into more open sky and wait. {% hint style="warning" %} A large accuracy radius (for example ±30 m or more) means the device is unsure where it is. Points captured in that state will pull the solve off or be rejected as outliers. Move to better sky view and let the fix improve before continuing. {% endhint %} *** ## Step 4: Capture localization points 1. Stand at your first point and **aim the camera at a clearly mapped part of the scene** (building facades, structures, and textured surfaces localize well; blank sky or featureless ground does not). 2. Hold still and tap the **capture** button (the location-pin icon at the bottom centre). 3. The app runs a VPS localization for that frame. On success you see **Localization Success** and the **Localized Locations** counter increases by one. 4. Follow the on-screen guidance. The app tracks how your points are spread and prompts you toward the gaps, for example *"Capture 4 more point(s), spread across the space"* and then *"Great spread. Capture a few more or tap Done to geo-reference the map."* 5. **Walk to a well-separated new spot, face a new direction, wait for GPS again, and capture the next point.** Repeat until you have covered the map.

Capture a point at each well-spread location. The counter tracks progress.

Once the spread is good, the app confirms and unlocks Done.

{% hint style="info" %} **How many points?** You need a minimum before **Done** unlocks (the on-screen prompt tells you how many remain). For outdoor maps we recommend capturing **8 to 12 well-spread points**. More good points, spread widely, give a more accurate and more robust result. If a capture fails, aim at a more clearly mapped area and try again. {% endhint %} If you make a mistake or the spread is poor, you can start over from the capture screen and re-capture your points. *** ## Step 5: Submit and geo-reference the map When the app reports a good spread and **Done** is enabled, tap **Done**. The app submits all captured control points to the server, which solves the geo-reference transform and returns a result you can review: * **Origin** : the solved latitude, longitude, and altitude of the map's local origin. * **Heading** : the map's rotation relative to true north, in degrees. On success, the map is geo-referenced and the solved origin and heading are stored for it automatically. You do not need to enter any coordinates by hand. {% hint style="success" %} A good result has a **low RMSE** and **most of your points used as inliers**. If several points are rejected or the RMSE is high, it usually means some points were captured with poor GPS or too close together. Re-capture with better sky view and wider spread. {% endhint %} *** ## After geo-referencing Once the map is geo-referenced, you can immediately use its geospatial features: * Provide a **GeoHint** (approximate GPS) in localization queries to narrow the search space and speed up localization. See [Geohint in Localization](/multiset/basics/localization/geohint-in-localization.md). * Blend outdoor and indoor experiences. See [Outdoor-Indoor Transitions with Multiset](/multiset/basics/georeferencing-maps/outdoor-indoor-transitions-with-multiset.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://multiset.gitbook.io/multiset/basics/georeferencing-maps/auto-georeference-map.md?ask=&goal= ``` `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.