uv101.

Guide · start here

How can a lens this small take pictures like these?

Written October 2026 by a beginner, researched with Claude. Sources at the end.

The main lens of the Seestar S30 Pro is 30 mm across, about the size of a large coin. For most of the history of astronomy, the rule was simple: a bigger lens or mirror shows you more. So how does a lens this small produce the Veil Nebula at the top of this site?

The short answer: the lens is the only thing about astrophotography that hasn't changed much. The camera chip behind it, the way pictures are combined, and the software that cleans them up have all improved enormously, and that is what this page explains.

The old rule: aperture is king

A telescope's aperture is the width of its main lens or mirror. It sets two things.

Light grasp. A lens collects light in proportion to its area, so doubling the width collects four times the light. Compared with a dark-adapted eye, about 7 mm wide:

ApertureExampleLight collected vs the eye
7 mmHuman eye1×
30 mmSeestar S30 Pro18×
80 mmTypical small refractor131×
200 mm8-inch amateur telescope816×
2,400 mmHubble Space Telescope117,600×

Resolution. Light spreads slightly as it passes through any opening, which limits the finest detail a telescope can separate. A common rule of thumb, the Dawes limit, is 116 divided by the aperture in millimetres, in arcseconds. For the S30 Pro that is about 3.9 arcseconds, against 0.6 for an 8-inch telescope.

So by the old rule, a 30 mm lens collects about 1/45th of the light of an 8-inch telescope and sees about one-sixth of the detail. Both statements are still true. What has changed is everything else.

From Galileo to the smart telescope

Why small optics work now

1. The camera chip catches almost every photon

A photographic plate recorded about 1% of the light that reached it, and the best astronomical emulsions managed only a few percent. The S30 Pro uses a Sony IMX585 sensor. In the matching ZWO astronomy camera, ZWO measures it at about 91% quantum efficiency: nine in every ten photons are recorded.

That alone is roughly a hundredfold gain over film. In effect, the Seestar's 30 mm lens with a modern sensor records about as much light as a film camera behind a lens around ten times wider.

2. Almost no noise from reading the chip

Every time a camera reads out an image it adds a little electronic noise. On older cameras that read noise was large, so you had to take long exposures to rise above it. The IMX585 adds as little as 0.7 electrons per pixel in its high-sensitivity mode, which switches on at gain 200. That is why our nights run at gain 200.

With so little read noise, a hundred 30-second frames are almost as good as one 50-minute exposure. Short frames are much more forgiving, too: a gust of wind, a passing plane or a tracking error spoils one frame instead of the whole night.

3. Stacking: time replaces aperture

The clarity of an image, its signal-to-noise ratio, grows with the square root of the total exposure time. Four times the exposure gives twice the clarity. A small lens collects less light per minute, but a computer-controlled telescope can collect for many hours, night after night, and stack everything together. Friday's 2.3 hours on the Veil, added to Tuesday's 46 minutes, is exactly this trade.

4. The optics and the pixels are matched

The S30 Pro's 160 mm focal length and 2.9 µm pixels give an image scale of 3.74 arcseconds per pixel, almost exactly its 3.9 arcsecond resolution limit. Neither the lens nor the camera is wasted on the other. From a British garden, the atmosphere usually blurs detail at a similar scale anyway, so a much bigger telescope would often be held back by the air above it.

5. Big nebulae need a wide field

For extended objects like nebulae, the brightness reaching each pixel depends on the focal ratio (focal length divided by aperture) and the pixel scale, not on aperture alone. At f/5.3, the S30 Pro is a fairly "fast" optic. And its 4° × 2.2° field fits the whole Veil, nearly 3° across, or the whole Andromeda Galaxy, about 3° long. A big telescope with a long focal length would need a mosaic of many panels to cover the same sky.

The honest limit: total light from a target still scales with aperture squared. Small galaxies, planets and fine structure are where big telescopes still win clearly. The Seestar shines on large, bright-enough targets with plenty of exposure time.

6. Filters that ignore streetlights

The S30 Pro's dual-band filter passes only two slices of colour: about 20 nm around hydrogen-alpha at 656.3 nm, and about 30 nm around oxygen-III at 500.7 nm. Old sodium and mercury streetlights fall outside both slices. White LEDs spread their light across the spectrum, so a little still leaks through, but most of it is blocked. That's what makes nebula photography possible from a town garden. It doesn't help with galaxies or star clusters, which shine in all colours.

7. Software that does what experts do by hand

The last piece is processing. Calibration with dark frames, aligning hundreds of frames, rejecting satellite trails, measuring star colours against the Gaia space telescope's catalogue, removing light-pollution gradients with AI, and AI noise reduction and sharpening are what turn the raw stack into a picture. Ten years ago most of these were specialist skills. Now they are tools anyone can run, and an AI can run them all night.

Teaching an AI an experienced imager's workflow

Experienced astrophotographers follow a well-worn routine built up over years of mistakes. We set out to have Claude follow the same routine, and to learn from mistakes the same way.

What an experienced imager doesHow Claude does it here
Checks what's above the trees and roofs tonightMeasured the garden's skyline with the telescope's own camera, then plans each target's window around it
Frames the target carefullyPlate-solves a test frame and nudges until the target is centred within a few arcminutes
Focuses on a star with a Bahtinov mask or focus curveRuns a V-curve autofocus on real stars at the start and every hour
Chooses gain and exposure from the camera's noiseMeasured that gain 80 was read-noise limited and moved to gain 200 and 20–30 s
Dithers between framesRandom guide pulses every five frames
Watches for cloud and throws away bad framesCounts stars in every frame, pauses on cloud, and drops frames that won't align
Shoots darks, calibrates, stacksTakes darks at the end of the night and runs Siril on the office PC
Calibrates colour against real starsPhotometric colour calibration from catalogue star colours
Stretches gently and avoids over-processingA masked, gentle stretch, learned after I called the first attempt "noisy as hell"

The loop that makes this work is feedback. When something goes wrong, the fix becomes code and the lesson is written into Claude's notes for next time. Examples so far: wrong star colours, a mirrored image, a noisy finish, the app holding the camera, cloud fooling the polar alignment. A human imager builds this experience over years. Here it accumulates in days, but it still needs a human looking at the pictures and saying what looks right.

Small telescopes doing real science

This isn't only about pretty pictures. Unistellar's network of small smart telescopes has contributed to published research. In November 2021, 43 observations from nine countries recorded a 16-hour transit of the exoplanet Kepler-167e, the longest-period planet whose transit had been caught from the ground (Perrocheau et al., 2022). For the Seestar there are, so far, mainly anecdotal reports of exoplanet transits and asteroid measurements.

Sources