Archival Image Calibrator: correcting exposure, tone and colour errors in digitised images against FADGI criteria
Year: 2026
Type: preprint
Abstract: Digitisation work to the FADGI four-star level is judged on the images, not the equipment, yet real capture sessions contain human errors: unevenly placed lights, exposure that changes between shots, and charts that drift from their reference values. We describe Archival Image Calibrator, an open-source Python tool that corrects tone, white balance and colour in TIFF images from a photographed reference target. A per-session mode fits per-channel monotone tone curves and a neutral-preserving root-polynomial colour matrix from one ColorChecker SG capture; a per-image mode registers a chart found inside each frame and fits that frame's own tone curve. Every correction is written to an auditable record. On a Hasselblad X2D II 100C setup with deliberately imperfect lighting, the corrected ColorChecker SG capture passed the FADGI four-star tests in Open DICE (mean DeltaE2000 1.82, 90th percentile 3.10), and per-image correction reduced held-out colour error on three Golden Thread frames to a mean DeltaE2000 of 1.4-1.7. The tool does not correct spatial non-uniformity of illumination; evidence is limited to one camera system and few frames.
Language: English
Subfield: Atomic and Molecular Physics, and Optics
Field: Physics and Astronomy
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7225733093