Best DSLR or ILC for Night Sky Imaging?

Started Nov 8, 2017 | Questions thread
rnclark Senior Member • Posts: 3,755
Re: Sony A7S

camera_nerd wrote:

sharkmelley wrote:

If you don't mind the large pixels then the Sony A7S is the best out there in terms of sensitivity in low light for two reasons:

1) The low read noise in each (large) pixel means it has the lowest read noise per unit area of sensor than any other camera I know of.

2) The camera is very effective at keeping sensor heat low during continuous use. This gives it the lowest thermal noise per unit area of sensor than any other camera I know of. Take a look at this graph of how dark current (which translates into thermal noise) rises during continuous 5 minute exposures:

These low noise levels give you far greater flexibility when choosing shutter speed and F-ratio.

The big downside with the Sony A7S is the star eater issue but that only happens in bulb mode and you are unlikely to use bulb mode for astro-landscapes. Many people don't notice the star eater issue in any case.

The A7SII uses the same sensor but because the sensor is image stabilised it is much more difficult to conduct the heat away, so the dark current and thermal noise rise faster. Also the star eater kicks in for exposures longer than 4 seconds. The latest firmware (FW 4.0) reduces some of the star eater effects in the green channel but it still kicks in at 4seconds and above.

In addition, if you set it to video mode you can see the Milky Way, aurorae and the brighter galaxies and nebulae in real time - especially if you reduce the frame rate to the slowest setting of 4 frames/sec.



I am new to DPreview forums and have just started experimenting with astrophotography. Just got a Sony a6000 and tried it with an adapted Minolta MD 50mm f/1.7 lens. This lens has bad coma, but it let me test the a6000 till I can get a better lens.

Last night in the wash behind my condo( I live in southern Arizona 25 miles south of Tucson) I tried exposures at f/2.0, ISO1000 and ISO1600, 10 secs, 15 secs, 20 secs. The best shots were ISO1000 for 10secs. In the longer exposures the sky was gray, not black.

Also at the previous new moon I tried my Nikon D7100 but it exhibits sensor glow starting at 10 second exposures.

I would like to have a FF camera, but as a retired PC tech on a fixed income, I have a budget. And so far my cost/benefit analysis has not favored going to FF, since I currently do not have any FF lenses.

What is the source of the Dark Current graph? That is the kind of information that I find very interesting.

Have spent a good bit of time on Photonstophotos. Great resource! Thank you Mr. Claff.

And thank you, sharkmelley for your information.


The data were measured by Mark. Mark does great work and I trust his results. But I do feel the presentation is deceptive and gives an incorrect impression of which camera is best. But I am not sure how to better show all these cameras in a better way because of complexities of different pixel sizes on the image.

Let's focus on just 2 cameras. The Canon 7D Mark II which plots in the upper group, and the Sony A7s which plots the lowest so appears the best.

A side note, Mark sent me a couple of his dark frames from the 7D2, and the camera shows significant banding that mine or other 7D2s I have seen data for do not have. You can see my data for the 7D2 here: My measurement of Mark's data indicates about a factor of 2 higher dark current than in my camera. (Disturbing that there is such variation in copies, but we'll ignore that for the moment.)

So marks plot normalizes the data to dark current per square micron. The pixel size of the A7s is 8.4 microns while the 7D2 is 4.09 microns, thus areas of 70.6 and 16.7 square microns, respectively. As I've discussed and demonstrated previously, a sharp lens (e.g. even modern f/1.4 lenses, like the Sigma Art 35 f/1.4 or Canon 35 f/1.4 II, will produce smaller stars and show finer detail in images made with smaller pixel size cameras. To first approximation a star will require 2x2 pixels. So the dark current contributing to the star for the A7S will have 4 pixels, or 4*70.6 * dark current, while the 7D2 will have 4*16.7 * dark current.

If we choose time point 9 on Mark's plot, the A7s is about 0.004 e/micron/sec and the 7D2 is 0.017 per second. So the dark contributing to the star is 4*70.6 * 0.004 = 1.13 e/sec and the 7D2 has 4*16.7 * 0.017 = 1.13 e/sec.

So we see they are very close.  Add the sky noise and we would see the 7D2 pulls ahead.  Then change to a lower noise 7D2 and it pulls ahead more.  The smaller pixels will show fainter stars, as well as smaller stars and more detail.

The point is not A7s versus 7D2, it is large versus small pixels and that the data need to be interpreted carefully in terms of real world imaging, and in practice results will be much closer than it seems by some sensor data, especially for nightscapes where exposure time is short relative to deep sky astrophotography.  In fact, for most nightscape photography, the data to use in Mark's plot is at the left edge where the cameras are bunched together.


Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow