ClarkVision.com

To Clarkvision.com Galleries
Home Galleries Articles Reviews Best Gear Science New About Contact

Characteristics of Best Digital Cameras and Lenses
for Nightscape, Astro, and Low Light Photography

by Roger N. Clark


Night and astro photography requires an optimum collection of light, because there is so little light available. Light collection places constraints on what to consider for nightscape and astrophotography. The principles also apply to low light photography in general.

The Night Photography Series:


Contents

Introduction
Lens Characteristics for Gathering the Most Light
Best Camera Characteristics for Low Light Photography
Crop versus Full Frame Cameras
Long Exposure Astrophotography Considerations
Discussion and Conclusions
References and Further Reading


All images, text and data on this site are copyrighted.
They may not be used except by written permission from Roger N. Clark.
All rights reserved.

If you find the information on this site useful, please support Clarkvision and make a donation (link below).


Introduction

First, as with all photography, the lens is the most important piece of equipment, but in low light photography, gathering light is critical, especially with stars and the night sky. It becomes even more so with astrophotography to record faint subjects like galaxies and nebulae. A nightscape image in natural light made with a large aperture lens is shown in Figure 1a. In this article, I'll describe the characteristics of lenses and cameras for making stunning night photos that include stars and the Milky Way. I'll also discuss digital cameras and lenses for making astrophotos of galaxies, nebulae and star clusters in the deep sky. My discussion will be limited to digital cameras and camera lenses, and not the many (and often many times more expensive) cooled CCD cameras, large telescopes, and exotic tracking systems needed to hold them. This article is about using wide angle to telephoto camera lenses for low light, nightscape, and astrophotography.


Figure 1a. Natural color image nightscape. Collecting light is the key to deep images like this one. The image is a mosaic using a Sigma 35 mm f/1.4 DG HSM lens on a full frame DSLR, a Canon 6D 20-megapixel digital camera. Gallery image with more information is here.

Lens Characteristics for Gathering the Most Light

There are common misconceptions regarding light gathering in photography. I'll first try and clarify light gathering by lenses as it impacts the choice of lenses for night photography.

Photographers are trained that more light gathering means a faster f-ratio. After all, exposure is directly related to the f-ratio. But f-ratio tells light density in the focal plane, not total light received from the subject. Light gathering from the subject is actually proportional to lens aperture area times exposure time. What this means is that for greater impact with night sky photography, buy the largest aperture lens you can afford. This means the fastest f/ratio in a given focal length. Note, this does not contradict my statement about f/ratio above. For example, a 15 mm f/2.8 lens has an aperture diameter of 15/2.8 = 5.4 mm, an aperture which is smaller than the dark-adapted human eye. A 35 mm f/2.8 lens has an aperture diameter of 35/2.8 = 12.5 mm and collects over 5 times, (12.5/5.4)2 = 5.3, as much light from the subject even though the f-ratios are the same. A 35 mm f/1.4 has an aperture diameter of 35/1.4 = 25.0 mm and collects (25/5.4)2 = 21 times more light than a 15 mm f/2.8 lens. That would be a huge impact in light gathering in night photography when light levels are so low.

Technical note. The lens aperture area times the subject angular area is called the Etendue. Etendue is is a key parameter used in designing optical systems, including cameras for spacecraft, aircraft, lab or field use. While the term is not known in the general photography community, Etendue describes the basic physics of light collection by an optical system and is key to distinguishing what is true from what are myths in the photography community. Etendue is also called the A Omega product, where A is the lens aperture area, and Omega is the subject angular area. For example, the Moon is one-half degree in diameter, so if the moon where the subject, the subject would be about 0.2 square degree, then Omega = 0.2 square degree. Etendue, combined with lens transmission, sensor quantum efficiency and exposure time can be used to measure absolute light levels with an imaging system, including digital cameras. Or it can be used to predict signal levels to set exposure times, e.g. by an orbiting spacecraft, or to compute the integration time to achieve a specific signal-to-noise ratio on a galaxy with your telephoto lens and consumer digital camera.

Etendue and Light Collection. Here is a demonstration of the above concepts. In forums about photography there are often arguments about exposure. There commonly seems to be confusion over brightness in an image and actual light collection. True light collection is what is important, how many photons were collected from a subject, because the signal-to-noise ratio from a subject in the scene is proportional to the square root of the amount of light collected. The noise we perceive in our digital camera images is almost entirely due to the noise in the light signal itself. Brightness in an image file depends on post sensor amplification (ISO). Only actual light collection (lens aperture area times exposure time) affects the light collection, not ISO or f-ratio. Examine Figure 1b and 1c for a demonstration. This is an extreme demonstration on purpose to illustrate light collection.

Say you want to make an image of the belt and sword region of the Constellation of Orion, including the Horsehead nebula, the Orion nebula and you have only 30 seconds to do it, and you can track to compensate for the Earth's rotation. You need the image for a web banner. Your fastest lens is a 35 mm f/1.4 and your next fastest lens is a 200 mm f/2.8. In 30 seconds, the 35 mm f/1.4 will certainly give a greater photographic exposure--a 2-stop advantage. This is illustrated in Figure 1b. The test is biased for the 35 mm f/1.4, with 2 stops greater light density and 1.5 times longer exposure time, the photographic exposure is certainly greater with the 35 mm image. But was more light collected from the subject: the area in the white box in Figure 1b? Does the greater photographic exposure record fainter stars and more nebulae? Photographic exposure is not the key metric. The key metric for light collection is the Etendue * exposure time.

Light collection is best described by Etendue * exposure time. Etendue is lens aperture area (technically the lens entrance pupil) times the subject solid angle. The subject can be a star, a galaxy, a bird in a tree, a person's face in a portrait, a person on stage in a play, or any other object we perceive in a scene. In the case of this example, the subject is the region in the box in Figure 1b. The frame is not a subject. The subject solid angle can also be a fixed angular size, like the box in Figure 1b, or a square degree or square arc-minute.


Figure 1b. Comparison of photographic exposure with a an f/1.4 lens and an f/2.8 lens. The box is the region of interest. There is a two stop difference in f-ratio and the exposure time with the slower lens is only 2/3 that of the faster lens. Which lens collected more light from the desired area?

From online discussions, it seems that many photographers would choose the 35 mm f/1.4 lens to make the desired image because of the greater exposure in the given time. But let's compute which lens and exposure time collects more actual light. We will ignore differences in lens transmission because these are small--a few to ten or so percent difference. If A = lens aperture area, Omega = subject solid angle, and T = exposure time, the light collected is A * Omega * T. Because the subject is the same with the two lenses, the relative light collected is simply lens aperture area times exposure time, A * T. Figure 1c, panels a and b give the values of A * T. and interestingly we see that the darker exposure (by more than 2 stops) actually collected more light according to the equations. NOTE: we do not need to know anything about sensor size, pixel size or number of pixels in the camera--those parameters are not part of the Etendue equation and are not needed.

If we sum the light from the 200 mm image into the same output pixels as in the 35 mm image, we see that there was actually so much light collected that the result is a totally saturated image (Figure 1c, panel c)! This a common problem in image processing, so instead of summing the light, and average is done. This prevents blowing out the image, and instead of raising the signal too high, averaging pushes the noise down. One can then stretch the image as desired (e.g. Figure 1d) and we can see more detail and fainter nebula and stars with the processed 200 mm "underexposed" image compared to the 35 mm f/1.4 image with a 2+ stops more exposure. This is because the 200 mm image collected significantly more light from the subject area.


Figure 1c. Illustration of photographic exposure compared to actual light collection. The 35 mm f/1.4 image show greater exposure (panel a) compared to the 200 mm lens (panel b). But if we sum the light from the 200 mm image to match the 35 mm image scale, we see that there is so much light in the 200 mm image that the signal gets saturated (panel c). Instead, we average pixels which reduces noise, enabling the image to be stretched and show more detail with fainter stars and fainter nebulae (panel d). The faintest stars detected with the 35 mm lens is 12.8 versus 14.3 for the 200 mm lens. The difference is exactly predicted Etendue * exposure time (panel e). The star in the lower right in panel e is indicated by an s and is also shown in panels a, b, and d for reference. The 35 mm f/1.4 lens as a 25 mm diameter aperture and the 200 mm f/2.8 lens has a 71.4 mm diameter aperture.

The ratio of the Etendue * exposure times in Figure 1c is 13.4 / 2.45 = 5.5. So the 200 mm image, despite over 2 stops less photographic exposure should show fainter stars by 5.5x, or 1.8 magnitudes fainter. Indeed, the 35 mm image shows stars to about magnitude 12.4 and the 200 mm image to about 14.6, or 1.8 magnitudes fainter (Figure 1c, panel e). Adding/averaging pixels together (this is called binning) reduces contrast on faint stars. The full resolution 200 mm image shows even fainter stars.

Summary: lens aperture area * exposure time describes the collection of light, not photographic exposure. Let's explore these implications further.

Real-World Examples. Examine the images in Figures 2a and 2b, made with 15 and 35 mm focal lengths with a 1.6x crop camera at f/2.8. The images illustrate 2 things. 1) One does not need really wide angle lenses (like 15 mm) as commonly cited as a requirement for Milky Way photography. Both 15 and 35 mm images make interesting nightscape images, and in my opinion, the image made with the larger aperture lens, the 35 mm f/2.8 is the more interesting image. 2) The larger aperture diameter lens (Figure 2b) collects more light. Full pixel crops comparing 15 mm f/2.8 versus 35 mm f/2.8 are shown in Figure 3a. Note the lower noise, more stars, and better detail in the 35 mm f/2.8 image. The 15 mm f/2.8 image collects too little light from the night sky in a 30-second exposure. The image, with so little light, is noisier.

The images in Figures 1-3 use tracking to keep the stars round. If the camera is on a fixed tripod then exposure times must be reduced as focal length increases, partially reducing the advantage of the longer focal length, larger aperture lens. But as I showed above, the 35 mm f/2.8 lens collects 5.2 times as much light. On a fixed tripod, we must reduce exposure proportional to the increase in focal length. Light gathering for the same f-ratio increases by the square of the focal length, so with the reduced exposure time for fixed tripods, we gain only as the ratio of the focal lengths. It is still a win. And because very wide angle lenses are not available in as fast f-ratios, one can gain more. For example, a 15 mm f/2.8 lens versus 35 mm f/1.4 lens (Figure 3b). The aperture area ratio is (25 / 5.4)2 = 21 times more light gathered. So if we can track, and keep exposure times the same (Figure 3b) we win big. If we must reduce exposure time by 15/35 = 0.43x, we only gain by 9.2x more light. That is still huge!

If you get a simple star tracker, then you don't need to shorten exposure times with increasing focal length, gaining more light. You can make a simple barn door tracker (see part 5 of this series) for a few dollars, or buy a commercial tracker, like an iOptron SkyTracker for a few hundred. Or simply stack 2 or 3 short exposures.

Now examine the images in Figure 3b and 3c, a comparison of images made with 35 mm f/1.4 and 15 mm f/2.8 lenses. It is clear that the 35 mm f/1.4 image is vastly superior.


Figure 2a. The Milky Way in natural color made with a 15 mm f/2.8 lens with a 1.6x crop sensor size, 30 second exposure, ISO 1600.


Figure 2b. The Milky Way in natural color made with a 35 mm f/2.8 lens with a 1.6x crop sensor size, 30 second exposure, ISO 1600.


Figure 3a. Full pixel view of images from a 35 mm f/2.8 lens (left) and 15 mm f/2.8 lens (right). Natural color. Of course the image scale is different, but note the noise in the brown star clouds. The noise is greater in the 15 mm lens because the smaller aperture diameter collects less light from the subject. The 35 mm lens also records more and fainter stars in the same exposure time.


Figure 3b. Full pixel view of images from a 35 mm f/1.4 lens (left) and 15 mm f/2.8 lens (right). Natural color. Of course the image scale is different, but note the noise in the brown star clouds. The noise is greater in the 15 mm lens because the smaller aperture diameter collects less light from the subject. The 35 mm lens also records more and fainter stars. The 35 mm lens collects over 21 times as much light in the same exposure time.

If you want tp make images of the Milky Way presented so that the subject is the same size, as in the Milky Way appears the same size on a print or screen, The image made with the longer focal length needs to be resampled and the larger aperture, longer focal length lens produces the better image. I show the effect of resampling the larger focal length lens to have the same amount of pixels on the subject in Figure 3c. Of course, for the same total field of view coverage, one would need to make a mosaic with the longer focal length. Prints made from the wide angle lens versus a mosaic with a large aperture longer focal length show a stunning difference (Figure 3c). Here we see the effect of noise on the subject, the Milky Way. The 15 mm f/2.8 image shows a lot of noise. The noise masks many faint stars, and the noise reduces contrast between the brighter dust clouds and the dark clouds in the Milky Way. The 35 mm f/2.8 image shows fainter stars than the 15 mm f/2.8 showing that you do not get the same amount of light with the same f/ratio. The 35 mm f/1.4 image shows very little noise, increased contrast and fainter stars and produces the overall best image of the 3 shown.


Figure 3c. The images from Figures 2a, 2b, 3a, 3b resampled to the same pixels on subject as the 15 mm image. Natural color. On the same scale, the larger aperture lens produces the better image with lower noise, finer detail, and fainter stars. The 15 mm image is so noisy, the noise looks like stars, and people often complain that their images show too many stars when it is really the noise in their images made with small apertures. The 35 mm f/1.4 image (right) is less busy, with less noise and the number of stars looks more balanced. The box in the right two panels indicates where the Signal-to-Noise ratio (S/N) measurement was made. We would expect an improvement of 2.3x and the measured improvement is 2.5x for the 35 mm f/2.8 lens over the 15 mm f/2.8 lens. The 35 mm f/1.4 image has a higher S/N by another factor of two and shows real detail at fine scales not seen in the other two images.

The Mosaic Advantage. As shown in Figure 3c, mosaicking with a larger aperture longer focal length lens has an advantage regarding noise and detail. But it goes further. Some images made with mosaics are difficult to impossible with wide angle lenses if one wants to record the same faint detail. For example, the image in Figure 1 covers 88 by 85 degrees and is a 31 frame mosaic made with a 35 mm f/1.4 lens on a full frame camera. A 14 mm lens, available in f/2.8 would almost cover the width (81 degrees). The Figure 1 image used 19 thirty second exposures for the sky and 4 positions on the land, stacking 3 120-second exposures (6 minutes for the land). A 14 mm lens has an aperture of 14/2.8 = 5.0 mm compared to the 25 mm aperture diameter for the 35 mm f/1.4 lens. The 35 mm lens collects 25 times the amount of light. On the sky, the 14 mm lens would need an exposure of 12.5 minutes. But the stars are moving, some rising on the left, and setting on the right, and stars are going behind and emerging from the mountain peaks. Thus one still needs multiple exposures. to assemble a reasonable view in time of the stars in the sky. For the land, a 6*25 = 150 minute exposure would be needed! Thus, the 14 mm image would take over 3 hours to gather the same amount of light from the scene. The 35 mm image can be done in under 1 hour. Further, I do the horizon line first to minimize star movement relative to the land, making final assembly relatively simpler. This image of the Maroon Bells with water reflection would be impossible to do with a wide angle lens like 14 mm f/2.8 and get the same signal-to-noise ratio as with a 35 mm f/1.4 lens. The long exposure times would mean one could not line up the stars in the sky with the reflection because multiple 12.5 minute exposures would be needed on both the sky and reflection and the stars move too much in this case. With wide angle lenses, one is forced to keep total exposure time lower, thus making a noisier image, fewer stars and lower impact.

Figure 4 shows examples of details in deep sky objects made with a 35 mm f/1.4 lens on a full frame DSLR (Canon 6D). Figure 4 shows full resolution detail you can get with a 35 mm f/1.4 lens and tracking/stacking several exposures. Detail includes spiral arms in galaxies, stars in star clusters and shapes of emission nebulae.


Figure 4. Full pixel resolution images made with a 35 mm f/1.4 lens and Canon 6D camera (6.4 micron pixels). Natural color. Note the dark lanes in M31, the spiral arms of M33, the blue dust in M45, individual stars in the Double Cluster, and the shapes of the Heart and Soul nebulae. Shorter focal length lenses do not resolve such details. The red background in some images is due to airglow and aurora. The images were made using a Sigma 35 mm f/1.4 DG HSM lens on a full frame DSLR, a Canon 6D 20-megapixel digital camera.
Gallery image with more information is here.

You can do nightscape images with longer focal lengths too, for example, as shown in Figure 5. Longer focal lengths, even with mosaics, mean a smaller field of view, and more precise planning of location and the times when interesting stars are above the landscape are required. But the results can have incredible detail. Note, the mountain in Figure 5 is the same mountain as in Figure 1.


Figure 5. Mosaic of images made with a Canon EF 100mm f/2 USM Lens lens on a 1.3x crop camera. Natural color. This illustrates one can make stunning nightscape images with any focal length, not just wide angle. The green sky is from airglow. Gallery image with more information is here.

Best Camera Characteristics for Low Light Photography

Summary Regarding Selecting a camera

Details

Contrary to popular internet opinion, the main factor, after lens aperture in long exposure photography, e.g. about a minute and longer, or even tens of seconds in warm environments, is noise from dark current. Popular internet opinion focuses on read noise, but read noise is insignificant in long exposures compared to noise from light pollution, airglow, and dark current, especially when used with lenses faster than about f/4. One can only beat light pollution in full color imaging by imaging the night sky far from cities, but dark current goes with the camera. Dark current is very camera dependent, and few reviews measure it. Dark current is measured in many reviews here on Clarkvision.

Low dark current is a key factor in long exposure low light photography. First some facts about dark current. Dark current doubles every few degrees increase in temperature. Typically, the doubling in CMOS sensors is every 5 to 6 degrees Centigrade. To be really precise with dark current subtraction (used in astrophotography), one needs the dark current to be measured at the same temperature to within a fraction of a degree. Noise from dark current is the square root of the dark current from the total exposure time (including stacking) and is independent of sub-exposure times. For example, if you make 50 1-minute exposures using a camera with 1 electron/second dark current, the noise from dark current in the stacked image is square root ( 50 exposures * 60 seconds per minute *1 minute) = 55 electrons.

Another big factor in image quality in nightscape and astrophoto images is banding in the camera. Banding is a fixed pattern usually horizontal and/or vertical. We are very sensitive to detecting banding so it becomes objectionable even when the peak-to-peak banding is 10 times smaller than random noise in the image. Some cameras have banding problems at some ISOs and not others. As ISO is increased, banding problems usually decrease. Banding is shown in reviews here on Clarkvision.

There is one camera that I have tested or seen data from other testing that stands well above the others: that is the Canon 7D Mark II digital camera (Figure 6, below). See the 7D Mark II review here and look at Figure 3 and Table 3 of the review and the corresponding discussion. Some recent Sony and Nikon numbers by people on dpreview indicate dark currents similar or slightly better than the other Canon cameras on Figure 6 below but I have not seen any data better than that for the 7D2.


Figure 6. Dark current for several cameras is compared as a function of internal camera temperature. Note the amazing improvement in low dark current with the Canon 7D Mark II digital camera. Sony sensor data are from commercial data sheets found online.

Newer model cameras from all manufacturers generally have better sensors, with lower noise (banding, read noise, dark current), so choose the latest model cameras for best results, especially if they have reviews that show if the camera has banding issues and a measurement of dark current.

Key new technology is called dark current suppression. Dark current suppression technology is hardware and part of the pixel design. It is not something your turn on or off in software. It is not long exposure noise reduction. It is not high ISO noise reduction. The hardware in the latest sensors use a 4-transistor circuit in the pixel. Some use 3 transistors. This is called 4T and 3T designs. What dark current suppression technology does is block the signal, but not the noise from dark current. The technology became established in circa 2008 digital camera models, and has been refined with newer models. What this means is in sensors that have the technology, we do not need to measure dark frames and subtract them in post processing. We no longer see amp glow in long exposure images (usually seen as pink blobs on the edges of frames). This technology allows a big simplification in post processing. If your camera shows amp glow, it would be a benefit to upgrade to a newer camera (models post circa 2008). The newest models generally have better refinements, meaning longer and longer exposures without the problems so common in older models. Figure 7 shows the major advance in technology that dark suppression and other sensor improvements have enabled in the last decade. If you have an older camera, pre 2010 and certainly pre 2008, an upgrade to a very recent model can be a big benefit. See Part 7b) On-Sensor Dark Current Suppression Technology of this series for more detail.


Figure 7. Comparison of technology. The image on the left was obtained in 2005 and on the right 2015. The image on the right collected 36% more light per pixel, so if technology were equal, the apparent noise difference would be only about 17%, barely noticeable. Obviously, the improved sensitivity and lower noise of modern cameras makes a huge difference. See the gallery image of the 2015 image here. The 2015 image was made with a Canon 7D Mark II 20-megapixel digital camera and 300 mm f/2.8 L IS II lens plus a Canon Extender EF 1.4X III giving 420 mm at f/4 and ISO 1600.

Another thing to consider in selecting cameras with low dark current (remember, the dark current suppression technology blocks the dark current level, but not the noise), is camera models that will dissipate heat more efficiently. The large massive pro cameras are at a disadvantage here. Their shear mass, generally faster electronics, more electronics (e.g. dual cpus), mean more heat and the heat that gets generated is harder to dissipate with all that mass. Lighter smaller cameras tend to dissipate heat better, and with slower electronics, generate less heat.

But regardless of new camera, you will see the biggest impact on image quality by getting quality lenses with larger apertures. For example, 24 mm f/1.4, 35 mm f/1.4, 50 mm f/1.4. The Sigma Art series are excellent. Rokinon/Samyang are cheaper but be aware people are finding they need to return multiple lenses before getting a good copy.

Crop versus Full Frame Cameras

Contrary to popular internet belief, larger sensor cameras have little to do with sensitivity. We often read on the internet that full frame cameras are more sensitive and that they are better at low light photography. This is a misunderstanding of the light gathering of lens and sensor. A larger format ENABLES one to use a larger lens. It is the lens that collects the light; the sensor is just a bucket to collect the light delivered by the lens (Figure 8).


Figure 8. Raindrops (blue) as an analogy to photons entering a camera. The buckets ("pixels") fill with water (blue), but the orange funnels (the lens) collect the rain and focus it onto the buckets (pixels). In this case the large end of the funnels have the same diameter so collect the same amount of rain per unit time (the exposure). The ONLY difference is the small bucket will fill up faster, but that is not a problem in low light situations. Small buckets are NOT a disadvantage. The funnel diameter controls how much rain goes into the bucket, not the size of the bucket. It is the same with cameras, lenses, and pixels. The angle "a" is the is the angular size of a pixel and is the same for both the large and small pixel in this example. In a camera, both cameras would get the same amount of light per pixel, show the same noise, have the same pixels on the subject, and have the same depth of field.

Example. For a given field of view, e.g. a 35 mm on a full frame sensor gives a 54.4 by 37.8 degree field of view. To get that same field of view on a 1.6x crop camera, one needs a (35 / 1.6 =) 21.87 mm focal length lens. Say the 35 mm lens was f/2.8, with an aperture diameter (35 / 2.8 ) = 12.5 mm. To collect the same amount of light with the 21.87 mm lens on the crop camera, it would need the same aperture diameter (technically called the entrance pupil) of 12.5 mm, making an f-ratio of 21.87 / 12.5 = 1.75. Typically, photographers keep the f-ratio constant, thus compare a 35 mm f/2.8 on the full frame to 21.87 mm f/2.8 on the crop. But a 21.87 mm f/2.8 lens has a smaller diameter aperture (21.87 / 2.8 =) 7.8 mm, and collects 2.5 times less light from the subject (12.5/7.8 squared)! The f-ratio tells light density, NOT total light from the subject. This concept seems particularly confusing to photographers trained that f/ratio tells about light. Yes, f-ratio tells light density thus exposure, but not total light from the subject. If you don't believe this, see Figures 9a, 9b and the text below, or if you want to learn more see part 1a of this series with Figures 4a, 4b, 5a, 5b and my series on Understanding Exposure.

Here is another way to look at the problem. Think of it this way: you have a full frame camera and after you take the image, you crop the image. You changed the field of view. You did not change the actual focal length. A crop sensor is just a smaller sensor--think of it as full frame pre-cropped so you don't have to crop in post processing. It does not change the lens attached in any way--the focal length is the same. The aperture is the same. The amount of light gathered within the frame is the same. The smaller sensor just means a smaller field of view. Why people think that changes sensitivity is surprising.

To understand effect of sensor size, try this analogy, we both go to a hardware store, You buy a 5-gallon bucket, I buy a 1-gallon bucket. We then go to a water faucet that is dribbling a low rate of water and fill our buckets for 30 seconds. How much water do we each have in our buckets? The amount of water in our buckets is not dependent on the bucket size unless one bucket overflows, and with the low rate of water from the faucet, neither bucket overfills (analogy to low light photography). The amount of water in the bucket is dependent entirely on the faucet delivering the water (analogy to the camera lens) and the length of time of the fill (the exposure time). The amount of water in the bucket has nothing to do with bucket capacity. We have the same amounts of water in our buckets. It is the same with pixels: the amount of light captured in the pixel is dependent on the lens delivering the light. There is one additional factor in cameras and lenses: the area the pixel covers. If we equalize the area, and thus pixels on subject, then using the same lens, same f/ratio, same exposure time, the light per pixel is the same, as in Figure 9a.

The lens delivers the light. The pixel is just a bucket. Below in Figures 9a, and 9b are comparisons of images made with a full frame and 1.6x crop cameras. If we believed the internet myth that larger sensors are more sensitive, one would expect to see fainter stars in the image made with the larger sensor and a noise difference between full frame and crop cameras to be the square root of the pixel areas, or sqrt(2.59) = 1.6, which would make the crop camera image noticeably noisier. Clearly, that is not the case in Figure 9a. Web sites that show a differences between crop and full frame cameras are typically changing the lens aperture area between cameras, thus the amount of light collected. If we keep the camera with small pixels at its full resolution and enlarge the image from the camera with the larger pixels, as in Figure 9b, we do see that the small pixels are noisier per pixel, but there are more pixels in the same area as a pixel from the larger pixel camera, thus the camera with smaller pixels show finer noise, and more detail. And because the pixels are smaller, the stars are smaller, and there is less contribution of noise from sky glow and dark current, so the camera with smaller pixels shows substantially fainter stars.


Figure 9a. Comparison of images made with a full frame and 1.6x crop cameras. The ratio of the pixel areas is 2.59. Focusing on the subject, not the pixel (after all what is the subject in a photo: the subject, or the pixel?), I binned the smaller pixels of the 1.6x crop camera to the same pixels on subject in the full frame. By resampling to the same pixel on the subject, we see that the signal-to-noise ratio (S/N) is essentially the same in the uniform areas (agreement to better than 1 percent), but the smaller pixels of the crop camera actually record fainter stars and more detail, even after resampling. Compare the unbinned images in Figure 9b.


Figure 9b. Comparison of images from two cameras--the same as in Figure 9a, but the crop sensor image is shown at full resolution, and the full frame camera enlarged 1.6x for comparison. The smaller pixels on the crop camera record fainter stars, over 70% fainter (0.6 magnitude). The noise per pixel is higher in the crop sensor camera, but there are more pixels, so the noise appears smaller. There is also more fine detail in the crop sensor camera image, thus providing better overall image quality despite technically having more noise per pixel. And as shown in Figure 9a, when the pixel angular areas are normalized, the noise is basically the same.

You can make great nightscapes and great astrophotos with both cropped sensors and full frame sensors. I have made nightscape and astrophoto images with both cropped and full frame cameras. My preferred astrophotography camera when I am doing many minutes of exposure is a camera with low dark current, regardless of sensor size. I pay less attention to sensor size in choosing a camera for a low light job, and pay more attention to sensor characteristics like low dark current and the lens that I will use.

The only other factor to consider regarding cameras is being able to change lenses. Very fast wide angle lenses are only available in single focal lengths (commonly known as prime lenses) with only a few rare exceptions. Most zoom lenses are not as fast and produce lower quality star images (again there are some exceptions). This generally means a DSLR or mirrorless camera with interchangeable lenses. Note that too much use of live view, whether mirrorless camera or DSLR heats the sensor increasing dark current noise. Thus I prefer DSLRs because I can frame and use the optical viewfinder to minimize sensor heating. Once a sensor heats up, it can take a half hour or more to cool back to ambient temperatures.

Long Exposure Astrophotography Considerations

Astrophotography with uncooled digital cameras has specific challenges. As discussed above, dark current can be significant, and when imaging in a low light pollution dark environment, dark current is usually the limiting factor in imaging faint objects. Second, the light from galaxies, nebulae and others interesting objects in the deep sky is very faint. That means the largest aperture are you can put to work is important. Dark current scales with pixel size, meaning larger pixels have more dark current than smaller pixels.

This leads to the non-intuitive concept of using the largest aperture to concentrate the light onto the smallest pixels!


Figure 10. The Whirlpool Galaxy, M51. This image was obtained with a Canon 7D Mark II 20-megapixel digital camera, a camera with small pixels and 300 mm f/2.8 L IS II lens plus a Canon Extender EF 1.4X III giving 420 mm at f/4, ISO 3200 and 2-arc-seconds per pixel. Gallery image and more information is here. Image scale is 2 arc-seconds per pixel.

Pixel Size. As I have already discussed, sensor size has little to do with collecting light from the subject. Similarly, pixel size also has little to do with light collection. The quantum efficiency (QE) per square mm is the same (for the same technology) whether a tiny sensor in a cell phone camera, or a full frame DSLR. This means if you concentrate the same amount of light onto a spot 10 microns square versus 5 microns square, the same proportion of light will be captured by the sensor. But because smaller pixels have less dark current, a system with smaller pixels can in practice perform better at low light long exposure photography.

For example, consider a 600 mm f/5.6 lens with a digital camera having 8 micron pixels. Call this System A. The angular size of one pixel is 2.75 arc-seconds (this is called the plate scale). The lens diameter is 600/5.6 = 107 mm. Consider a second system, System B, with a 300 mm f/2.8 lens and a digital camera with 4-micron pixels. System B has the same lens diameter (300/2.8=) 107 mm, and the same plate scale of 2.75 arc-seconds per pixel. The lens in Systems A and B will each deliver the same amount of light to the pixel. Assuming the image quality of the two lenses is the same, both will produce equal images in the same exposure time, EXCEPT in long exposures where noise from dark current is a factor (deep space astrophotography). This is where the smaller pixels have an advantage.

The next thing to consider in astrophotography is plate scale for imaging different deep space objects. A few objects are quite large, like the Great Nebula in Orion, M31, which is about double the diameter of the full Moon, the the Andromeda galaxy, M31, which is over 8 times wider than the full Moon. There are also some larger structures of the Milky Way that can be imaged by shorter focal length lenses. There are also many smaller objects, including galaxies and nebulae, which require finer resolution to get much detail in them. But atmospheric turbulence limits resolution in most cases except in rare circumstances. Long exposure astrophotography is often limited to about 2 arc-seconds. The image in Figure 10 is at 2-arc-seconds per pixel and if you follow the link to the gallery image, you will see a larger image at 1.5 arc-seconds per pixel. See; Astrophotography and Focal Length: What Can You Image with Various Lenses? for more examples.

For deep sky astrophotography, I usually choose lens and sensor to give between 1 and 4 arc-seconds per pixel depending on the size of the object. For nightscapes, it can be more about field of view and larger structures, so plate scales in the tens of arc-seconds per pixel are chosen. A 200 mm lens on a camera with 4 micron pixels gives 4.1 arc-seconds per pixel, and 800 mm gives 1 arc-second per pixel. See plate scale for scales with specific cameras and lenses. See Lens Field of View for data on camera sensor sizes and lenses.

Discussion and Conclusions

Putting the above information together, the ideal astrophotography system is very fast lenses feeding cameras with small pixels for a given plate scale. As with nightscape photography, where lenses like 24 mm f/1.4 and 35 mm f/1.4 are ideal, moving up the angular resolution range, 50 mm f/1.4, 85 mm f/1.4, 100 mm f/2, 135 mm f/2, 200 mm f/2 or f/2.8, 300 mm f/2.8, 400 mm f/2.8, 500 mm f/4 feeding digital cameras with pixel sizes around 4 microns are ideal.

Wide angle lenses, like 15 mm f/2.8 can make for easy night sky images. But the small apertures mean little light is gathered. A 15 mm f/2.8 has a smaller aperture diameter (5.4 mm) than the dark adapted human eye (about 7 mm). The little light gathered by such lenses results in a lot of noise in the typical 20 to 30 second exposures typical done by nightscape photographers. That noise is confused with stars leading people to think they have too many stars in their images. Larger aperture lenses, like 24 mm f/1.4, 35 mm f/1.4 collect many times more light, producing images with less noise. The less noise shows faint stars better, making for a better balance of stars and Milky Way clouds (see Figure 3c). Mosaics made with these larger aperture lenses make cleaner images with more detail, higher contrast and greater impact.

Another take-home message from this article and the physics: use the largest aperture lens/telescope to collect the most light. Efficiency may mean doing a mosaic depending on available resources of lens aperture size, focal lengths and fields of view. One can compromise on the lens for simplicity (e.g. ultra-wide lens and single short exposure versus mosaic with a larger aperture lens), but that simplicity reduces light collection, increasing apparent noise and resolution. This is true for daytime as well as nighttime photography--the physics does not change, just in daytime photography there is generally more light so the compromise may result in an acceptable image.

See Recommended Cameras and My Gear List for Photography for specific cameras and lenses that I use, and watch here for recommendations of the best cameras and lenses for night and astrophotography for Canon, Nikon and other manufacturers.

Technical. The technical term for the above concept of lens aperture collects the light and the field of view of the pixel is called the Etendue. It is the fundamental property of imaging systems. See my series on Understanding Exposure for more information.


If you find the information on this site useful, please support Clarkvision and make a donation (link below).


References and Further Reading

Clarkvision.com Nightscapes Gallery.

Clarkvision.com Astrophoto Gallery.


The Night Photography Series:


Home Galleries Articles Reviews Best Gear Science New About Contact

http://www.clarkvision.com/articles/characteristics-of-best-cameras-and-lenses-for-nightscape-astro-photography

First Published March 14, 2016
Last updated March 19, 2023