1. Choose stable control points
Prefer features that are identifiable in both datasets: building-footprint corners, road intersections, bridges, walls, hydraulic structures or distinctive parcel geometry. Avoid features likely to have moved.
2. Distribute GCPs across the raster
Three GCPs are mathematically sufficient for an affine transformation, but not for assessing quality. For a historical sheet, 6–15 well-distributed points usually provide a much stronger diagnostic.
3. Choose the transformation model
Affine transformation handles translation, rotation, scale and shear. Projective transformation (homography) can compensate for perspective in a photographed map. Thin Plate Spline (TPS) provides a smooth local warp for non-uniform paper or survey distortions.
4. Inspect residuals and RMS error
GeoCalque reports an RMS error and per-GCP residuals. These values describe internal fit; they are not a guarantee of absolute positional accuracy. Always inspect the map between control points.
5. GIS interoperability
The GIS tools panel can export and import GCPs as CSV or GeoJSON, display residual vectors directly on the reference map, show cursor coordinates in EPSG:4326 or EPSG:3857, and import/export affine world files. Custom raster basemaps can also be added as XYZ, WMS or Web Mercator WMTS sources.