01Why the 500 rule is wrong on a modern camera
The 500 rule says: divide 500 by your full-frame equivalent focal length, and that is your longest exposure in seconds before stars trail. On a 24mm lens that gives about 21 seconds. It is easy to remember and it comes from 35mm film, where grain was coarse enough to hide a fair amount of movement, and where nobody was going to inspect a negative at 100% on a 4K monitor.
It ignores three things that matter enormously today. It ignores pixel pitch: a star can drift across a 6-micron photosite without visibly smearing, but the same drift crosses three of the 3.7-micron photosites on a 61MP body, and that is a visible streak. It ignores aperture, which changes how tightly a star is rendered in the first place: a wide-open fast lens puts more energy into a smaller point, so movement shows sooner. And it ignores where in the sky you are pointing, which changes how fast stars move across the frame by a factor of three or more.
02What NPF actually calculates
The NPF rule takes all three. Its accurate form is t = k × (16.856 × N + 0.0997 × f + 13.713 × p) / (f × cos δ), where N is the aperture, f the focal length in millimetres, p the pixel pitch in microns and δ the declination of what you are pointing at. The k factor is your tolerance: 1 for genuinely pin-sharp stars at 100%, 2 or 3 if the image is destined for a screen at normal sizes.
On a 24MP full-frame body at 24mm f/2.8 it lands close to the 500 rule. On a 61MP body it typically gives you half as long, and the difference is real: shoot at the 500-rule time on that body and the stars are small ovals when you zoom in.
03The answer is not a longer exposure, it is more exposures
The reason people cling to the 500 rule is that a shorter shutter means a darker frame, and pushing ISO to compensate means more noise. Stacking solves this properly. Take ten or twenty frames at the NPF time and average them in software, and the random noise cancels while the real signal reinforces. You end up with the cleanliness of a much longer exposure and stars that are still points.
The "frames to stack" figure above tells you how many exposures at your NPF time add up to four minutes of total light, which is a sensible target for a Milky Way frame from a dark site. It is more work at capture and far less compromise in the result, and it is what essentially every clean night-sky image you admire was made from.
04Two practical notes the calculator cannot give you
Focus is the real failure mode. More astro frames are ruined by focus than by trailing. Autofocus will not work on a star. Switch to manual, use live view at maximum magnification on the brightest star you can find, focus until it is the smallest point possible, then tape the focus ring: it drifts when the lens cools through the night.
Coma matters more than the last stop. Most fast wide lenses render stars near the corners as small wings rather than points when shot wide open. Stopping down one stop from maximum usually cleans that up dramatically, and costs you far less than it appears once you are stacking. Test your own lens on a clear night before you drive somewhere dark.
05Frequently asked questions
500 divided by your full-frame equivalent focal length, in seconds. On a 24mm full-frame lens, about 21 seconds. It comes from 35mm film and is too generous for modern high-resolution sensors inspected at 100%.
It accounts for aperture, pixel pitch and declination, all of which the 500 rule ignores. On a 45MP body it typically halves the allowed exposure, and it is correct to do so.
NPF if you inspect at 100% or print large; the 500 rule is fine for web-sized output. Best of all: shoot at the NPF time and stack more frames to recover the light.
Yes, by a factor of three or more. Stars at the celestial equator move the full 15 arcseconds per second; near Polaris they barely move. The Milky Way core at -29° declination moves at about 87% of the maximum rate.