I spent three years chasing the shadow. My first total eclipse — Svalbard, March 2015 — lasted two minutes and forty seconds, and I fumbled every frame. No amount of rehearsal on the partial phases prepares you for the instant the photosphere vanishes and the corona emerges: a silver-white halo impossibly delicate against a sky that has no business being that dark at midday.
The problem with photographing the invisible
The corona's dynamic range spans roughly fourteen stops — from the inner ring, bright enough to damage an unprotected sensor, to the outermost equatorial streamers, which register barely above the noise floor of any commercial camera. No single exposure captures it. The accepted solution is a bracketed sequence: seven to eleven frames, each separated by one stop, composited later with high-dynamic-range software. But stacking introduces artifacts at the lunar limb, where the sharp edge of the Moon meets the soft scatter of photospheric light, and every compositing algorithm struggles to resolve that boundary without haloing.