On the evening of August 12, 2026, the sun dropped toward the western horizon in a deep 99.34% partial eclipse. By shooting linear 14-bit RAWs through a homemade solar filter with zero sensor clipping, this photographic sequence became a calibrated instrument to measure solar limb darkening and atmospheric extinction.
Solar altitude, crescent tilt, and obscuration change dramatically depending on your exact geographic position. Explore how the eclipse appeared across Spain and Europe, or enter your own coordinates.
Why does the crescent tilt? As the sun sets, the celestial equator cuts into the local horizon at an angle equal to (90° - Latitude). In Gredos (40.35°N), the solar axis is tilted ~48° relative to the horizon, rotating the lunar bite from a high-afternoon 'horn' to a horizontal sunset 'smile'.
Step through the 58 centrable eclipse frames and 7 sunset coda frames. Every frame is registered with sub-pixel circle fitting and tone-mapped in linear light.
A 1D light curve cannot reveal which direction the moon crossed the sun. However, on any registered solar disc, the area-weighted first moment (centroid) of the lit crescent points strictly opposite the moon.
Because the lit crescent is the solar disc minus the circular overlap, its area centroid lies exactly on the line connecting the Sun and Moon centres:
By tracking this centroid across 58 frames, we reconstruct the moon's 2D trajectory vector with R² > 0.999 linearity, independently confirming the 0.051 R_sun closest approach distance.
Because the sensor never saturated (brightest photosphere sat 2.8 stops below clipping), the raw pixel counts are strictly proportional to photon flux, allowing real solar physics measurements.
When looking at the centre of the sun, our line of sight reaches deeper into hotter layers (6400 K). Toward the limb, we only see higher, cooler layers (4500 K). Because Planck's law causes hot gas to emit much more blue light, the solar limb darkens much faster in blue wavelengths than in red.
How consumer photographic hardware was transformed into a high-precision scientific instrument through sub-pixel geometry, sidereal drift modeling, and linear photometric calibration.
Direct conversion of Sony ILCE-6400 RGGB Bayer CFA data (B0 = 512, saturation at 16383). No camera tone curves, sharpening, or sRGB gamma applied, preserving strict photon flux linearity.
Algebraic Kåsa-Taubin circle fitting on the uneclipsed solar limb. Determines the exact optical center (x0, y0) and solar radius (R = 238.43 px) with residual RMS < 0.04 px.
28 manual tripod re-aims were decomposed into a continuous piecewise-linear celestial drift model at 3.635 px/s, recovering absolute temporal alignment (R² > 0.9998).
A sub-pixel threshold offset of δ = 0.818 px was calibrated on uneclipsed frames, correcting a +0.42% perimeter-to-area systematic bias on thin crescent measurements.
Möbius tone-shoulder compression and 3 calibrated grading LUTs: Warm Gold (solar atmosphere), True Solar Neutral (5778 K blackbody), and Monochrome (pure photon density).
Create astronomical sequence posters where distance along the axis is strictly proportional to physical time (x ∝ t). Renders real linear RAW image crops across all layout modes.
How consumer photographic gear was turned into a high-precision astronomical sensor.
Formulations derived from first principles for clock synchronization, celestial mechanics, and radiative transfer.
Local contact times $T$ are computed from the Besselian fundamental plane $(x, y, d, \mu, l_1, l_2)$:
Within each continuous tracking block $k$, the solar center in sensor pixels $(x_s, y_s)$ evolves as:
Linear sensor DNs are mapped to display luminances without clipping highlights:
Topocentric solar/lunar coordinates, contact times, and duration tables across Spain are calibrated against official Besselian elements published by the Instituto Geográfico Nacional (eclipses.ign.es).