loading…
—splat
Δx—Δy—Δz—mm
skull az—el—d—
fov—R—α—
— fps
S capture · H clean · C view · R photo · drag orbit · ⇧ pan · wheel dolly · W A S D Q E
Built from runs/ and captures/. Save a version to keep a dated copy.
To observe the skull depicted by Holbein, one must stand at a specific vantage point within the Renaissance collection at the National Gallery. This optimal viewing position can be accurately reconstructed from the panel through precise geometrical analysis, achieving millimeter-level accuracy as demonstrated by Idols of the Cave. However, the objective of this study is to utilize a predictive regression model to perform reverse engineering of the position that most closely approximates the “imaginary position” from which Holbein was situated during the creation of the painting.
The model is SHARP, Apple’s single-image reconstruction network. From one photograph it builds the room in the painting as about 1.2 million Gaussian splats, a cloud of small coloured ellipsoids the camera can move through. The photograph carries no camera data, so the model assumes a 30 mm lens, and every metre it reports follows from that guess. The study repeats the reconstruction at lenses from 5 to 200 mm and converts each into panel millimetres, so the position the model favours can be set beside the one geometry gives.