Your astrophotography copilot
Are more megapixels always better?
Camera ads sell the megapixel count. For deep-sky photos, it is one of the least useful numbers on the box. This guide covers what matters instead: pixel size, pixel scale, noise and what it looks like, and why a cooled astronomy camera beats a DSLR on a faint target.
No. What decides the image is how much light each pixel collects, how much sky each pixel covers, and how much noise you start with. Megapixels only tell you how many pixels there are. A 9 megapixel cooled camera on the right telescope can match or beat a 60 megapixel DSLR on the same target.
Think of each pixel as a bucket in the rain, and the light from the sky as the rain. A sensor is a fixed-size sheet of buckets. Cutting the sheet into more, smaller buckets does not catch any more rain overall. It only shares the same rain between more buckets, so each one fills less. That is the whole story of megapixels.
- More megapixels on the same sensor means smaller pixels. A 24 and a 60 megapixel full-frame sensor are the same size. The 60 megapixel one has pixels more than a third narrower.
- A pixel catches light in proportion to its area. Pixels half as wide catch a quarter as much light each.
- The whole sensor still collects the same light. Shrink the picture back down, or bin the pixels, and the extra noise averages out. A high-megapixel camera is not simply worse. It trades noise per pixel for a chance at more detail.
One sensor, one star, one sky. This is the same small patch of the sensor, zoomed in, cut into bigger or smaller pixels. A brighter square means that pixel caught more light.
- Pixels in this patch
- 96
- Light per pixel
- 25%
- Pixels across the star
- 2.9
The total light landing on this patch is the same in every case. It is just shared between more or fewer pixels.
The star covers about 2 to 4 pixels across. That records the star's real shape without wasting light.
So, bigger or smaller?
Neither is better on its own. The best pixel size is the one that matches the sharpness of your stars. Try the demo above: with a sharp sky, medium or small pixels capture the star well. With a soft sky, the small pixels only record the blur.
Bigger pixels
- Each pixel catches more light, so the picture is smoother in each frame.
- Forgiving: stars stay sharp even with soft seeing or small tracking errors.
- Best for wide fields, short focal lengths, big faint nebulae and the Milky Way.
- Smaller files and faster processing.
- The cost: finer detail is lost if your telescope and sky could have delivered it.
Smaller pixels
- Can record finer detail, but only if stars are already sharp.
- Each pixel catches less light, so each one is noisier.
- Best for small targets at long focal lengths, in steady air with accurate tracking.
- Bigger files, slower processing, more demanding of focus and guiding.
- The cost: if the sky is soft, you pay in noise for detail that was never there.
Aim for a typical star to span about 2 to 3 pixels across. For most amateur setups under ordinary seeing, that works out to about 1 to 2 arcseconds per pixel. If your pixels are much smaller, bin them or shrink the final image. If they are much bigger, stars turn blocky, which is fine for wide shots but not for close-ups.
Pick a DSLR and an astro camera and put them behind the same telescope. Notice that the camera with more megapixels often has the smaller pixels, and so catches less light in each one.
| A | B | |
|---|---|---|
| Megapixels | 24.2 MP | 26.1 MP |
| Pixel size | 5.93 µm | 3.76 µm |
| Sensor size | 35.8 × 23.8 mm | 23.5 × 15.7 mm |
| Cooling | None | Regulated cooling |
| One 16-bit raw frame | ~48 MB | ~52 MB |
| Pixel scale | 3.06″/px | 1.94″/px |
| Field of view | 5.12° × 3.41° | 3.37° × 2.25° |
With the same telescope, a pixel catches light in proportion to its area, so a pixel twice as wide catches four times as much.
A: Wide. Wide-field imaging. Stars may look slightly blocky at full size, but you'll capture big areas of sky.
B: A good range. Roughly 1 to 2.5 arcseconds per pixel suits most backyard skies and tracking mounts.
Sizes and pixel sizes are approximate, from the makers’ published figures. Pixel scale is the sky each pixel covers, and light per pixel ignores differences in sensor efficiency, so treat the bars as a guide to the effect of pixel size only.
The pixel scale is how much sky one pixel covers: 206.265 × the pixel size in micrometers ÷ the focal length in millimeters. A small refractor with a focal length of about 290 mm and a 3.76 µm camera gives about 2.7 arcseconds per pixel, which is good for a wide field. The field-of-view calculator does the sum for any setup you save.
| Pixel scale | What it means |
|---|---|
| Under about 0.7 arcsec per pixel | Oversampled for almost everyone. Needs a large telescope, excellent seeing and very good guiding. |
| About 1 to 2 arcsec per pixel | A sweet spot for ordinary seeing and amateur telescopes. Stars are sharp and each pixel catches plenty of light. |
| About 2 to 4 arcsec per pixel | Fine for wide fields and big nebulae. Stars look a little blocky when you zoom in, and a lens or small refractor gives you this. |
| Over about 4 arcsec per pixel | Undersampled. Good for the Milky Way and very wide shots, not for fine detail. |
These ranges are rules of thumb. A very fine pixel scale is not a mistake. Software binning, or shrinking the stacked image, turns small pixels back into larger ones.
| DSLR or mirrorless | Astro camera | |
|---|---|---|
| Temperature | Uncooled. The sensor warms up as it runs, so noise changes through the night. | Regulated cooling holds the sensor at a set temperature, so noise is lower and steady. |
| Infrared and red light | A stock filter blocks much of the deep red hydrogen light that nebulae glow with, unless the camera is modified. | Made to pass it, and often with a built-in or added IR-cut for the visible stars. |
| Color or mono | Color only, with a filter pattern over the pixels. | Color or monochrome, and mono works with filters for narrowband imaging. |
| Processing in the camera | Some models apply noise handling or star-eating filters to the raw data. Many can be switched off. | Raw, unprocessed data, which stacks cleanly. |
| Control | Buttons and an intervalometer, simple and familiar. Also good for daytime photos. | Usually run from a laptop or a controller. No screen or buttons. |
| Price | Often already owned, or cheap used. | A dedicated camera, usually more expensive than a used DSLR of similar size. |
Both can make fine images. The astro camera’s advantages are mostly about noise and consistency, which is where the next two sections go.
Noise is any random variation that is not part of the target. Different kinds look different, and they have different cures, so it helps to tell them apart.
Shot noise (photon noise)
Cause: Light arrives as individual photons, and counting them is random. The noise grows with the square root of the signal.
What it looks like: A fine, even grain across the whole frame, strongest on faint areas.
What helps: Only more light helps: a longer total exposure, a faster telescope, or a darker sky. Stacking more frames cuts it by the square root of the number of frames.
Read noise
Cause: Each time the camera reads the sensor, the electronics add a small random amount to every pixel.
What it looks like: A grain that is worst in short exposures of a dark sky, and sometimes with faint stripes.
What helps: Expose long enough that the sky signal rises well above it, and stack. Newer low-read-noise cameras help a lot.
Thermal noise (dark current)
Cause: Heat makes electrons spill into the pixels with no light at all. The amount roughly doubles for every 5 to 7 degrees Celsius of warming, depending on the sensor, and it grows with exposure time.
What it looks like: A brighter, grainier background, and a scatter of bright speckles called hot pixels.
What helps: Cool the sensor. The fixed part (the hot pixels) can also be subtracted with dark frames.
Fixed-pattern noise and amp glow
Cause: Each pixel has its own bias and dark current, and some sensors warm up near their electronics, so the pattern is the same in every frame.
What it looks like: Hot pixels, stripes, or a warm glow at an edge or corner.
What helps: Calibration. Darks, bias and flats remove the pattern, and dithering stops it from stacking up in one place.
Color noise
Cause: On a color camera, each pixel sees only one color and the picture is rebuilt from neighbors, so noise in one color shows up as a different tint than its neighbors.
What it looks like: Blotches or speckles of green, magenta and red in the dark areas, once the image is stretched.
What helps: More data, then color noise reduction in processing. It is usually easier to remove than the brightness grain.
A simulation, stretched the same way each time. It shows how noise adds up, not any one real camera.
Buried in noise. The nebula is hard to see.
Heat from the sensor adds 10.8 electrons of dark signal to every pixel in each 60 second frame, and every one of those electrons also carries random noise.
- Drag the temperature up: the grain, the bright specks (hot pixels) and the glow at the left edge all grow.
- Stack more frames: the grain falls with the square root of the number, but the hot pixels and glow stay.
- Subtract a master dark: the specks and glow go, though the grain from the heat stays.
- Make the pixels smaller: each one catches less light, so the same sky looks noisier pixel by pixel.
How it works
A thermoelectric (Peltier) cooler sits behind the sensor and pumps heat away to a fan or heat sink. Many cameras manage a drop of 30 to 40 degrees Celsius below the air temperature, and a regulator holds it at a set point such as minus 10.
Why it cuts noise so much
Dark current roughly doubles every 5 to 7 degrees. Cooling from 25 to minus 10 degrees is 35 degrees, which cuts the dark signal by something like 30 to 100 times, depending on the sensor. With far fewer stray electrons, there is far less noise and far fewer hot pixels.
Why a steady temperature matters
Dark frames only work if they match the lights, and the match depends on temperature. A regulated sensor sits at the same temperature all night, so one set of darks can be reused for months. An uncooled camera drifts, so each session needs its own.
What it does not fix
Cooling does not reduce shot noise, which comes from the light itself, and it does not make a short exposure of a faint target clean. It removes the heat-related noise so that light is the only thing left limiting you.
Use it well
Pick one set point and stay with it, since a different temperature means different darks. Cool down and warm up gradually so the sensor and window are not shocked, and use the dew heater, because a cold window collects moisture fast.
The numbers here are typical figures, and real sensors differ. Check your camera’s own documentation for its cooling range and dark current. See the calibration frames guide for taking darks at the same temperature as your lights.
Wide Milky Way and big nebulae with a lens
A DSLR or mirrorless with large pixels (5 µm and up) and a fast lens is excellent, and you probably own one already.
Wide-field deep sky with a small refractor
A cooled color camera with 3.7 to 4.6 µm pixels. You get 2 to 3 arcsec per pixel, which suits a 300 to 400 mm telescope.
Galaxies and small nebulae at long focal length
Smaller pixels (2.4 to 3.8 µm) on a longer telescope, and only if seeing and guiding are good. Binning gives you the large-pixel option back.
Starting out with minimal fuss
A smart telescope, or a modest cooled camera. Resist chasing megapixels, since a 9 megapixel sensor is plenty.
Myths to leave behind
- More megapixels means a better astro image. The result depends on pixel scale, light and noise, not the count.
- A big sensor always beats a small one. A big sensor covers more sky, but it needs a telescope that can fill it with sharp stars.
- Noise is something the camera adds that you can only hide later. Most of it comes from the light itself, which more exposure time fixes.
- Cooling makes any exposure clean. It removes heat noise, and shot noise and read noise remain.
- DSLRs cannot do deep-sky work. They can, and many great images come from them, just with more noise and more calibration work than a cooled camera.
Before spending more on a sensor, spend time on the things that move the needle more: total exposure time, a dark site, accurate focus, good tracking, and the histogram and gain settings.
- Are more megapixels better for astrophotography?
- Not by themselves. More megapixels on the same sensor means smaller pixels, and a smaller pixel catches less light, so each one is noisier. Extra detail only helps if your telescope, seeing and tracking can deliver it. For most amateur setups, pixels that give 1 to 2 arcseconds per pixel are the sweet spot.
- What is a good pixel size for astrophotography?
- There is no single best size. What matters is the pixel scale, the sky each pixel covers, which depends on pixel size and focal length. Aim for about 1 to 2 arcseconds per pixel for detail, and a larger scale for wide fields. Pixels between 2.4 and 5 micrometers are typical in astro cameras.
- What is pixel scale?
- The patch of sky covered by one pixel, in arcseconds. It equals 206.265 times the pixel size in micrometers, divided by the focal length in millimeters. The field-of-view calculator works it out for your setup.
- Why is my astro image noisy?
- Mostly because there is not enough light yet. Noise falls with the square root of the total exposure time, so four times the data halves it. Heat, short exposures, a bright sky and small pixels all make it worse, and calibration frames and cooling help.
- Why do astronomy cameras have cooling?
- Heat creates dark current, extra electrons that look like noise and hot pixels. A cooled sensor has a lot less of it, and it stays at a steady temperature, so dark frames match your lights for months. That is the biggest practical advantage over an uncooled DSLR.
- Can I get a cooled camera effect with a DSLR?
- Not exactly. You can shoot on cold nights, use shorter exposures and take darks at the same temperature as your lights. Some people fit cooling to a DSLR, but a dedicated cooled camera is simpler.
- Does binning help?
- Binning combines neighboring pixels into one, so a 2 by 2 bin acts like a pixel twice as wide with about four times the area. It cuts the resolution and the file size, and improves the signal per pixel. You can also do the same later by shrinking the stacked image in software.
Compare your own gear in the FOV calculator, or read about darks, flats and bias.