Focusing at Night: Five Ways to Nail Focus on Stars

Night photography’s harshest lesson is usually handed out in the morning: you put the card in the computer, open three hundred frames that were four hours of work, and see that the stars aren’t pinpoints but small bloated balls. The framing is good, the exposure is right, the noise is manageable — and all of it is rubbish, because focus was off by a few microns. Every other setting of the night can be salvaged; focus cannot.

The good news: this is a problem technique solves completely. The bad news is that the camera will be no help at all.

Why autofocus gives up at night

Autofocus systems, whether they use phase detection or contrast detection, need light and an edge to work: the system looks for a transition in the scene and tries to make it as sharp as possible. Starlight is badly insufficient for that job — the number of photons reaching the sensor is far below the threshold at which the AF algorithm can say “there’s an edge here.”

The result is familiar: the lens runs from one end of its travel to the other, hesitates a couple of times, and finally stops without confirming. None of the focus modes — AF-S, AF-C, eye detection, whatever — changes this picture. Some modern bodies have dedicated “starlight AF” modes and they genuinely work; but they too need a bright star, and at best they match manual focus. So the actual skill is still yours.

The lie of the infinity mark

Relying on the ∞ mark printed on the lens is night photography’s most common and most expensive shortcut. It fails for three separate reasons:

• Most modern lenses turn past infinity. This isn’t a manufacturing defect but a deliberate tolerance: to leave room for thermal expansion and production variance, autofocus lenses extend their focus travel a few degrees beyond the ∞ mark. Turning the ring until it stops takes you past infinity, and the stars scatter.
• The mark itself is approximate. The precision of a screen-printed line is orders of magnitude coarser than the real focus tolerance of a 20mm lens at f/1.4.
• Focus drifts through the night. As the air cools, lens elements and barrel contract; focus that was perfect at 10 pm can be visibly off by 3 am, once the temperature has dropped ten degrees. This is called thermal drift, and it’s more pronounced at longer focal lengths.

Five methods

These are ordered roughly by usefulness in the field, not by precision. The first method is enough for most nights; the rest come in when conditions get worse.

1. Live View + the brightest star (the gold standard)

This is the right way to do it, and it takes ten minutes to learn:

1. Mount the camera on a tripod, switch to manual focus, and open the aperture wide — a wide aperture both brightens the screen and narrows focus tolerance, making the error visible.
2. Turn on Live View and push ISO temporarily to the ceiling (12800–25600). The image will be noisy; it doesn’t matter, this frame won’t be taken.
3. Put the brightest star in the frame (or a planet if there is one — Jupiter and Saturn are perfect for this) in the middle of the screen.
4. Zoom the screen magnification all the way in: 100%, more if possible.
5. Turn the focus ring very slowly. You’ll watch the star grow and shrink. Correct focus is the point where the star looks smallest and hardest.
6. The critical detail: the only way to know you’ve found the right point is to pass it. Shrinking, shrinking, smallest… and growing again. Come back and stop at that smallest point. Approaching from one direction and saying “this looks about right” almost always means stopping slightly early.

2. Set it in daylight, then tape it

The most robust way to avoid fiddling in the dark is to finish the job in daylight. During the day, autofocus on something at least 200–300 metres away (a mountain ridge, a distant building, the horizon), then switch autofocus off and tape the focus ring. Don’t touch the lens until nightfall.

The method is especially valuable if you’re heading to unfamiliar terrain: you won’t be wrestling with focus in the dark on top of everything else. Its one condition is that you keep using the same lens at the same focal length. It also doesn’t fully solve thermal drift — it’s worth one verification around midnight.

3. Focus on a distant light

If the stars are behind cloud or moonlight is washing out the screen, an artificial light source on the horizon will do: a house lamp on the opposite slope, the red light of a distant tower, even a friend’s flashlight carried 300 metres out and pointed back at you.

The rule: the light must be beyond the hyperfocal distance. On wide-angle lenses that’s usually 30–50 metres, but in practice aiming for anything past 200 metres is safe in every condition. The further the better — hyperfocal logic works in reverse here: the closer you get to infinity, the larger your margin for error, which means you can relax.

4. Focus peaking

On mirrorless bodies, focus peaking marks in-focus edges with coloured speckles. For night work it is not enough on its own — on lights as small and faint as stars, peaking either never triggers or triggers falsely. But used together with 100% magnification it’s a good second opinion: set sensitivity to maximum, and at correct focus you’ll see speckles shimmer on the brighter stars.

Don’t gather frames trusting peaking alone; use it as a confirmation.

5. A Bahtinov mask (for precision work)

A cheap plastic or cardboard disc that mounts in front of the lens, with slots cut at particular angles. Pointed at a bright star, diffraction produces a three-spike pattern: two fixed spikes forming an X, and a moving line through the middle. When focus is exactly right, the middle line bisects the X perfectly symmetrically.

It removes entirely the subjectivity of hunting for the “smallest dot” by eye. It’s more precision than a wide-angle astro landscape needs, but at 85mm and above — or for moon photography — it makes a real difference. They cost a few dollars, and you can cut one yourself.

Verification: a single test frame

Believing you’ve found focus isn’t enough; before you start gathering frames for the night, demand proof.

Knowing this distinction matters: a focus error is identical everywhere in the frame, while an optical aberration is clean near the centre and degrades toward the corners. If the central star is bloated too, the problem is focus.

Holding focus through the night

Finding focus is half the job; not touching it is the other half.

• Tape it. Wrap a piece of masking tape (gaffer tape is better) so that it locks the focus ring to the barrel. Nudging the ring while reaching for the lens in the dark happens far more often than you’d think.
• Decouple autofocus from the shutter. Set up back-button AF in the body menu, or switch AF off entirely. If a half-press sends the lens hunting for focus, the rest of the night is in the bin.
• Turn image stabilisation off. On a tripod-mounted camera, the image stabilisation system can hunt for vibration that isn’t there and produce softness. For long night exposures, definitely off.
• Verify every hour. Thermal drift is real. Take a test frame at the top of each hour and zoom to 100%; that 20-second check prevents a three-hour loss.
• Don’t touch the zoom. If you want to change the framing, turn the tripod head, not the focal length.

The foreground problem

Pure-sky frames gather statistics; a photograph starts with an anchor in the foreground. But here you’re up against physics: a 20mm lens focused at infinity at f/2.8 can’t render a rock three metres away sharply. Depth of field at night is already thin because of the wide aperture.

There are three honest solutions:

• Push the foreground further away. Easiest: move the anchor out to 15–20 metres. At that distance, hyperfocal keeps infinity and the foreground acceptably sharp at once.
• Two frames, two focus points. Shoot the sky frame focused at infinity; then, without disturbing the framing, focus on the foreground and take a second frame, painting it with moonlight or a low-intensity flashlight. Blend the two in editing — a standard, legitimate method in astro landscape.
• Settle for a silhouette. If the foreground is a dark silhouette anyway, nobody notices slight softness. The least laborious and usually the most elegant solution.

Common mistakes

• Turning the ring until it stops. You end up past infinity. The ∞ mark is an approximate signpost, not a destination.
• Approaching from one direction and stopping. You only know you’ve found correct focus by passing it and coming back.
• Focusing on a star in the corner of the frame. Corners are where optical aberrations concentrate; you can’t judge focus by looking there. Always pick a star near the centre.
• Being fooled by screen brightness. In the dark, your eyes are dilated, so the screen looks brighter than it is and focus feels “good enough.” Base your judgement on 100% magnification, not on the screen.
• Never verifying through the night. Set-and-forget is the right strategy on interval nights like the Perseids — but one check frame per hour is the legitimate exception to that rule.
• Solving focus but not the sky. A perfectly focused frame still shows an empty sky under light pollution. Focus is the first condition, not the only one.

Focus is the easiest skill to learn in night photography and the most expensive one to neglect. Once you’ve done it right, it settles into muscle memory; after that it’s the first ten minutes of every setup and never crosses your mind again. If you’ve also settled the basic night settings, all that’s left is the generosity of the sky. One more night puts the same skill to the test at long focal lengths: photographing a lunar eclipse means setting focus once and holding it for hours.