How a survey is made

A survey begins on site and ends in a file that can be measured. Between those two points the building is photographed from the air and from the ground and the two sets are resolved into one model.

Capture

Aerial. A UAV flies the monument in overlapping passes: orbits at several altitudes for the superstructure and nadir passes for anything a person on the ground cannot see. A vimana, a roof and the upper courses of a wall are all out of reach of a tape measure; the aerial pass records them from the one position they can be seen from.

Terrestrial. The same building is photographed at close range: plinth mouldings, niches, joints, damage and every surface that the aerial pass sees at a grazing angle and therefore resolves poorly.

Neither set is sufficient alone. Aerial capture loses the underside of every overhang; terrestrial capture loses everything above reach. The model is built from both.

Processing

Images are aligned and resolved into a dense point cloud and a textured mesh in Agisoft Metashape. The mesh is the measurable object: every dimension we publish is taken from it and not from the drawing derived from it.

Tolerance

Accuracy is a property of a specific survey, not of a studio. It depends on ground control, image count, capture geometry and conditions on the day. Vaayil therefore states tolerance per record, in that record’s documentation panel, rather than advertising a single figure across the archive.

Where a survey’s accuracy has not been established (where ground control was not placed or where the provenance of an early survey is still being recovered from project files), the record says so in What we do not know and no figure is published. An unstated tolerance is honest. An assumed one is not.

Access

Nothing is flown or entered without permission. Where a permission reference exists it is published with the record; where a building is in active worship, survey is arranged around use, not the other way round. See Ethics and access.