# University of Ilorin — data acquisition and reference policy

## Reference location and model extent

The current imported OSM extract identifies the campus at **8°29'9.22452"N, 4°40'28.4988"E** (`8.4858957, 4.6745830`). This named university node is the project reference/control point, not surveyed control. Geometry retains its separate bLOS M origin at `8.48199987411499, 4.671600103378296`; use that origin for placemark projection. See [coordinate provenance](COORDINATE-PROVENANCE.md).

The active context AOI (retained on 10 September 2026) is:

| Edge | Decimal degrees |
| --- | --- |
| South | `8.4715` |
| West | `4.6600` |
| North | `8.4955` |
| East | `4.6840` |

Once this loads smoothly on a phone, divide the wider campus into named zones and create separate GLBs or optional detail layers.

## Recommended source hierarchy

| Source | Use it for | Reliability in this project | Include in a published GLB? |
| --- | --- | --- | --- |
| OpenStreetMap | Footprints, roads, paths, water, land-use, building levels/roof tags | Good starting context; tags can be incomplete | Yes, with attribution and ODbL compliance |
| ArcGIS World Imagery / elevation | Visual comparison, broad ground shape, date/source checking | Useful reference; imagery and elevation resolution vary by place | Normally no imagery; check the precise service licence and credit requirements |
| Google Earth Pro | Visual inspection, historical comparison, manually created placemarks and measurement notes | Reference only; not survey control | No Google imagery, 3D mesh, or captured tiles in the GLB |
| Your permitted ground photos and basic measurements | Facades, entrances, signs, roof form, relative dimensions | Best source for recognisable detail without drone/RTK | Yes, if you own or have permission to use the images |
| RTK/total-station/drone survey later | Dimensions, elevations, alignment, measurement claims | Survey-grade only when a proper control workflow is used | Yes, after formal QA and provenance recording |

## 1. OpenStreetMap through bLOS M — the model backbone

Use OSM for the exported *geometry foundation*, not merely as a background map. bLOS M can import buildings, water, vegetation areas, roads, paths, railways, and terrain; it uses OSM height, floor-count, building-part, and supported roof tags when they exist. [bLOS M documentation](https://github.com/vvoovv/blosm/wiki/Documentation)

1. In Blender, open the `bLOS M` panel in the 3D View sidebar.
2. Choose **terrain** first and select the initial AOI around the current import AOI. Rename the resulting object `TERRAIN_BASE`.
3. Then choose **OpenStreetMap** and use exactly the same AOI. Import to a collection called `CONTEXT_OSM`.
4. Inspect the model before styling it. Treat a rectangular OSM extrusion as an unverified mass, not a confirmed building form.
5. Move the landmark buildings that you will refine into `HERO_BUILDINGS`; retain the original OSM object in a hidden `SOURCE_OSM` collection for traceability.
6. In the finished Three.js viewer, retain the visible credit `© OpenStreetMap contributors` plus a link to the OSM copyright page. OSM data is licensed under ODbL and requires attribution. [OSM copyright and attribution](https://www.openstreetmap.org/copyright)

**Do not trace Google imagery into OSM.** OSM's own guidance warns contributors not to add data from copyrighted sources without explicit permission.

## 2. ArcGIS — high-quality private reference and terrain comparison

bLOS M explicitly supports **ArcGIS Satellite** as an image overlay when you provide an ArcGIS access token. Use it in Blender to compare an OSM footprint with visible roofs, roads, landscape edges, and recent construction. [bLOS M image-overlay instructions](https://github.com/vvoovv/blosm/wiki/Documentation)

Recommended procedure:

1. Create or use an authorised ArcGIS account/token and add it in bLOS M preferences.
2. Import terrain, then select **image overlay → ArcGIS Satellite** and project it onto `TERRAIN_BASE`.
3. In ArcGIS Map Viewer or ArcGIS Pro, inspect the imagery information at the site to record provider, resolution, and capture date where available. World Imagery identifies source metadata and its sources vary by location. [Esri World Imagery metadata](https://doc.arcgis.com/en/data-appliance/latest/imagery-elevation/world-imagery.htm)
4. Use the overlay to correct manually modelled hero buildings, then hide it before export.
5. Build the public ground from procedural materials, your own imagery, or imagery you explicitly have rights to redistribute. If a specific ArcGIS service licence permits publication, retain its required provider credits exactly.

ArcGIS elevation is a reasonable broad terrain check; it is **not** a substitute for RTK, a total station, or a surveyed campus DEM. bLOS M's standard terrain import is approximately 30 m resolution, so use it for slope and drainage context, not for floor elevations or building height validation.

## 3. Google Earth Pro — useful, but keep it reference-only

Google Earth Pro is valuable for looking at historical imagery, confirming whether a building existed in a particular period, navigating the current import AOI, and recording your own placemarks. Its imagery is collected at different dates and may be a mosaic; 3D imagery does not have a single reliable collection date. [How Google Earth imagery is collected](https://support.google.com/earth/answer/6327779)

Use it this way:

1. Search for `8.4858957, 4.6745830`.
2. Create placemarks for landmarks and record a note for each: name, visual source/date, confidence, and whether you photographed it yourself.
3. Use the ruler only for *relative modelling checks*—for example, whether one wing is roughly twice another—not as a surveyed dimension.
4. Use the historical imagery slider to flag buildings that may be absent from older imagery.
5. Save only your placemarks/notes as KML or KMZ if that helps you organise the work. Google Earth Pro supports KML and GeoJSON-related workflows. [Google Earth Pro release notes](https://support.google.com/earth/answer/40901)

Do **not** export, scrape, capture, or convert Google Earth 3D buildings/photogrammetry into an OBJ, GLB, texture atlas, or substitute 3D city model. Google's published guidance prohibits using Earth output or third-party capture tools to reconstruct 3D models or similar content in conflict with its terms. [Google Earth guidance](https://support.google.com/earth/thread/235706120/is-it-possible-to-export-3d-models-from-google-earth-pro-to-work-with-in-3d-programs?hl=en)

## 4. Minimum field collection that materially improves the model

You do not need drone or RTK data to produce a credible visual campus explorer. For each hero building, capture:

- One straight-on photo of each main facade; avoid people and number plates where possible.
- One photo of each entrance, sign, and roof edge visible from the ground.
- A building name confirmed by campus signage or an authorised campus source.
- Floor count and any visible roof type.
- One simple real-world reference measurement when permitted, such as a doorway width, bay width, or a measured facade length.
- The capture date and photographer/permission status.

Assign each exported object one of these confidence levels:

| Confidence | Definition |
| --- | --- |
| `Verified` | Footprint and name checked using permitted observations; key dimensions measured or surveyed |
| `Photo-verified` | Footprint/form checked against photos; height or some detail estimated |
| `OSM-derived` | Imported from OSM with no dedicated review |
| `Illustrative` | Intentional placeholder; never imply it represents the real building |

## 5. Three.js delivery policy

Your production GLB should contain modelled geometry, original/properly licensed textures, and object metadata—not web-map tiles. Use `display_name`, `description`, `type`, `confidence`, `source`, and `reference_date` custom properties in Blender. The existing viewer reads them when a visitor clicks a building.

Add a source panel or a footer in the final viewer with:

```text
Approximate student navigation aid — the map is not to scale.
Base vector context © OpenStreetMap contributors (ODbL).
Building detail: campus observation and permitted reference imagery; dates vary by building.
```

If a later survey, drone mission, or authoritative GIS data is acquired, retain the original source date, coordinate system, control method, and horizontal/vertical accuracy before replacing these approximate layers.

Thirty image-derived site labels are documented in [image label provenance](IMAGE-LABEL-PROVENANCE.md). They retain their own image evidence and uncertainty; no Google image pixels or traced building outlines are included in either GLB.