On a fixed tripod the sky turns under your camera, and every star draws a short line during the exposure. Keep the exposure short enough and the line hides inside the star's own blur, so it still looks like a point. The question is where that limit sits, and the answer you were probably taught is wrong for modern cameras.
The 500 rule and why it fails
The 500 rule says the longest exposure is 500 divided by the focal length, times the crop factor for smaller sensors. At 24 mm on full frame that gives 20.8 seconds. It came from the film era, when grain was much coarser than the pixels on a modern sensor, so a short trail disappeared into the grain.
On a 24 megapixel full-frame camera, with pixels about 6 microns across, 20.8 seconds at 24 mm moves a star near the celestial equator by about 6 pixels. That is a visible streak at full size, and it softens stars even in a scaled-down print. On a 45 or 60 megapixel body the same exposure trails across even more pixels.
The NPF rule
Frédéric Michaud of the Société Astronomique du Havre published a better rule in 2010. It accounts for three things the 500 rule ignores: pixel size, aperture and where in the sky you point.
t = k x (16.856 N + 0.0997 f + 13.713 p) / (f x cos(dec)) N f-number f focal length, mm p pixel size, µm dec declination of the stars k 2 for normal viewing, 1 for stars that stay round at 100%
The aperture term is there because a lens at f/2.8 blurs a star over a larger spot than at f/1.4, which hides a little more movement. The declination term is there because stars near the celestial pole move in small circles and barely shift, while stars on the celestial equator move fastest.
Worked numbers
A 24 megapixel full-frame camera (6.0 µm pixels) and a 24 mm lens, with the standard k = 2 that PhotoPills and most apps show:
| Setup | NPF limit | Pixel-sharp (k = 1) |
|---|---|---|
| 24 mm f/1.8, celestial equator | 9.6 s | 4.8 s |
| 24 mm f/2.8, celestial equator | 11.0 s | 5.5 s |
| 24 mm f/1.8, Milky Way core (dec -29°) | 11.0 s | 5.5 s |
| 24 mm f/2.8, Milky Way core | 12.6 s | 6.3 s |
| 14 mm f/2.8, Milky Way core | 21.4 s | 10.7 s |
Compare that with the 500 rule's 20.8 seconds at 24 mm. The NPF limit is about half as long, and that half is the difference between pinpoint stars and short dashes.
Where does ISO come in?
It does not change the limit at all. A star moves the same distance across the sensor at ISO 800 as at ISO 6400. ISO only changes how bright the image looks, which is why it matters once the exposure is fixed.
Brightness scales with exposure time times ISO, divided by the f-number squared. A common dark-sky starting point for the Milky Way is 20 seconds at f/2.8 and ISO 3200. At the NPF limit of 11 seconds at f/1.8, the same ISO 3200 is about 0.4 stops brighter than that, so ISO 2500 matches it. At 12.6 seconds and f/2.8 you are about two thirds of a stop darker, and ISO 5000 makes up the difference. Under a brighter sky you need less of all of it.
Getting more than the limit allows
- Stack. Twenty frames at the NPF limit, aligned and stacked, carry far less noise than one frame. Sequator and Starry Landscape Stacker handle the foreground separately from the sky.
- Go wider. The limit grows as focal length shrinks. At f/2.8 on the Milky Way core, a 14 mm lens allows 21.4 seconds against 12.6 at 24 mm.
- Use a star tracker. A small tracking mount turns the camera with the sky, so exposures of one to three minutes stay sharp. The foreground blurs instead, so most people shoot it separately.
One more thing that softens stars
The NPF rule assumes a perfect lens. Fast wide lenses at full aperture often show coma and astigmatism in the corners, which turns stars into small wings or crosses regardless of exposure. Stopping down by a third to a full stop usually cleans this up, at the cost of light. If corner stars look bad at every exposure length, the lens is the cause, not the sky's motion.