Milky Way Exposures by the Numbers: The 500 Rule, the NPF Rule and Where to Focus

Daytime exposure gives you three free dials. At night, one of them is fixed before you start. The sky rotates about 15 degrees an hour, so every star in the frame is crossing the sensor while the shutter is open. Leave it open too long and points become short streaks. Those streaks are hard to see on the camera’s rear screen and very obvious on a monitor. That limit decides the shutter speed, and the rest of a night exposure is a question of finding the other stops from aperture and ISO.

This post works through that in order. First the maximum shutter time, then what the sensor format does to it, then where the remaining light comes from, then where to focus. One thing to say at the start: no tool on this site calculates the 500 rule or the NPF rule. Both are short enough to work out by hand, and both are below. The tools come in afterwards: the Crop Factor calculator gives you the number the 500 rule needs, the Exposure Triangle Calculator moves stops around once the shutter is set, and the Hyperfocal Distance calculator tells you how much foreground you keep when focused for stars.

The 500 rule is quick and too generous on modern sensors

The old rule of thumb divides 500 by the full-frame-equivalent focal length:

max exposure (s) = 500 ÷ (focal length × crop factor)

A 14 mm lens on full frame gives 500 ÷ 14 = 35.7 s. The rule comes from 35 mm film viewed as modest prints, where a short trail disappeared into the grain and the print size. It doesn’t consider resolution at all. A 45-megapixel sensor gets the same 35.7 seconds as a 12-megapixel one, even though its pixels are much smaller and show the same trail at a larger size when you zoom in. That’s why people moved to the 400 or 300 rule as resolution went up. Those are still guesses, though, just more cautious ones.

The NPF rule includes aperture and pixel size

The NPF rule, worked out by Frédéric Michaud for the Société Astronomique de France and made popular by the PhotoPills app, replaces the guess with the factors that actually decide whether a trail shows. Its simplified form is:

max exposure (s) = (35 × N + 30 × p) ÷ f

Here N is the f-number, p is pixel pitch in microns, and f is the actual focal length in millimetres, not the equivalent. Pixel pitch is sensor width divided by image width in pixels: 36 mm ÷ 6,000 px = 6.0 µm for a 24-megapixel full-frame body, and 36 ÷ 8,256 = 4.36 µm for a 45-megapixel one.

Here are the two rules side by side at f/2.8 (f/2 for the Micro Four Thirds lens):

Setup	Pixel pitch	500 rule	NPF rule
Full frame 24 MP, 14 mm f/2.8	6.0 µm	35.7 s	19.9 s
Full frame 45 MP, 14 mm f/2.8	4.36 µm	35.7 s	16.3 s
APS-C 24 MP (1.5×), 16 mm f/2.8	3.92 µm	20.8 s	13.5 s
Micro Four Thirds 20 MP (2×), 12 mm f/2	3.34 µm	20.8 s	14.2 s

On the 24 MP full-frame body the 500 rule allows nearly twice the NPF time, and at 100% view that difference shows up as obvious trails. The simplified formula assumes the worst case, which is stars on the celestial equator, because they move fastest. The full NPF formula divides by the cosine of the declination you’re aiming at. The galactic core sits at about −29°, which allows about 14% more time (19.9 s becomes about 22.7 s on the first row). The simple version leaves you that margin for free.

Crop factor shortens the limit, and it isn’t framing that causes it

In the 500 rule the crop factor sits in the denominator. The same 14 mm lens on a 1.5× APS-C body drops from 35.7 to 500 ÷ 21 = 23.8 s. The Crop Factor & Equivalent Focal Length Calculator gives you that equivalent focal length (21 mm here) along with the crop-scaled aperture. Only use the focal length for this. As the crop factor guide explains, the equivalent aperture describes depth of field and total light, not exposure, so it has nothing to do with star trails. It does matter for noise, though. At the same framing and f-number a smaller sensor collects less light in total, which is why f/2.8 on a crop sensor isn’t full-frame f/2.8 when you compare how clean two night shots look.

The reason a smaller sensor trails sooner isn’t that the frame looks tighter. A lens projects a star’s movement as the same number of microns on any sensor behind it. The difference is that a small sensor gets enlarged more to reach the same viewing size, and its photosites are usually smaller, so those microns cover more pixels. That’s why the NPF rule doesn’t need a crop factor at all. Pixel pitch already carries it. It also explains a result in the table: at matched framing (24 mm-equivalent on APS-C versus 14 mm on full frame), the APS-C body still gets the shortest time, 13.5 s. It needs a longer real focal length (16 mm) to frame the same view, and its 3.92 µm pixels are the smallest of the three f/2.8 setups.

A fixed shutter means aperture and ISO supply every remaining stop

Once the shutter is capped, the exposure triangle has two dials left. The exposure triangle guide explains why a stop is a stop on any dial. At night that means every stop you don’t get from aperture has to come from ISO.

Aperture: usually wide open, sometimes one third of a stop down. Fast wide lenses often show coma (stars at the edges drawn out into little wings) and heavy vignetting at maximum aperture. Closing down by a third or two thirds of a stop usually cleans that up. The NPF rule makes the cost slightly smaller than you’d expect. Stopping the 14 mm from f/2.8 to f/4 loses a full stop of light, but the NPF limit rises from 19.9 s to 22.9 s, which gets back about 0.2 of a stop. That’s because N is in the numerator: at smaller apertures each star is already a slightly larger blur, so a little movement hides inside it.

ISO: whatever is left. Many recent sensors are close to ISO-invariant above a moderate base setting, meaning raising ISO in camera gives much the same noise as brightening a lower-ISO file later, although that varies between models. A common starting point for the Milky Way core from a properly dark site is ISO 3200 at f/2.8 for whatever the NPF rule allows. Treat that as somewhere to start and adjust from your histogram, not as a fixed recipe.

Using the exposure calculator at night. Here’s what the Exposure Triangle Calculator does, from its code. ISO goes up to 102,400 and aperture down to f/0.95, both in third stops. The shutter list, though, uses full stops only and tops out at 30 s: 30, 15, 8, 4, 2 and 1 second. There’s no 13 s or 20 s option, so round your NPF figure down to the nearest listed value and think of the tool as a stop-counting aid, not a list of real camera settings. It shows EV at ISO 100 from aperture and shutter, and a running total of stops relative to its ISO 100 · f/8 · 1/125 default.

A common way to work in the field is to take a quick high-ISO test shot to check the composition, then convert it to the real exposure. Suppose the test frame is ISO 25,600 at f/2.8 for 4 s. Lock aperture in the tool and change the shutter from 4 s to 15 s. That’s log₂(15 ÷ 4) = 1.9 stops brighter, so the tool brings ISO down by the same amount and lands on ISO 6400 (the exact value, 6,827, isn’t on its list). The tool reports the setting as +19.9 stops brighter than its daylight default, which gives a sense of how far below daylight this kind of shooting is.

Focus on the stars at infinity, not at the hyperfocal distance

The hyperfocal guide shows that focusing at the hyperfocal distance gives you everything from half that distance to infinity “acceptably sharp”. For a landscape, that’s the most efficient focus point. For stars it isn’t, because in the hyperfocal setup infinity is right at the far edge of the depth of field. Distant objects are blurred to exactly the circle of confusion. On full frame the tool uses 0.030 mm for that, which is five pixels on a 6.0 µm sensor. A mountain blurred by five pixels still looks like a mountain. A star blurred by five pixels becomes a soft, dim disc, and since a star’s whole job in the picture is to be a sharp point, it loses brightness and colour as well.

So focus on the stars themselves. Magnify live view on a bright star or planet and turn the focus ring until it’s as small as it gets. Don’t trust the infinity mark on the lens, because many lenses focus slightly past infinity on purpose to allow for temperature changes. Once it’s set, tape the ring.

Then check how much foreground is still sharp. When you focus at infinity, the near limit of sharpness is the hyperfocal distance itself, not half of it. With the 14 mm at f/2.8 on full frame, the Hyperfocal Distance & Depth of Field Calculator gives 2.35 m. Enter its maximum focus distance of 1,000 m to stand in for infinity, and the near limit comes out at about 2.34 m. Anything beyond that is fine at the 0.030 mm standard. With the 16 mm on APS-C (0.020 mm) the hyperfocal distance is 4.59 m, so the foreground has to sit further back.

If you want a stricter standard that matches viewing at 100%, the tool doesn’t accept a custom circle of confusion, but its 1-inch option uses 0.011 mm. Choose it while keeping your real focal length and aperture, and the 14 mm f/2.8 hyperfocal distance becomes 6.38 m. That’s a more honest answer for a large print. If a rock or tree closer than that needs to be sharp, take a second frame focused on it and blend the two later. Stopping down to bring it in would also cost you shutter time or ISO.

Before you head out

Work out your NPF limit at home for the lenses you’ll take, using pixel pitch from your camera’s spec sheet, and write the numbers down. In the field, use the Exposure Triangle Calculator to convert a high-ISO test frame into the final exposure, the Crop Factor & Equivalent Focal Length Calculator if you’re working with the 500 rule on a smaller sensor, and the Hyperfocal Distance & Depth of Field Calculator to check whether your foreground is beyond the near limit once you’re focused at infinity.