Menu
#186. Low-light long exposures with the Sony A7r
By pascaljappy | Review
Since my initial review of the Sony A7r, I have been asked by several readers to perform long exposure tests in low light, with 2 goals in mind :
While the first condition seemed very unlikely (my impression – at this early point – is that exposures from 1/30 to 1/50 are most affected by vibrations) I set up the camera on a Feisol carbon fiber tripod and Arca Swiss monoball P0 for 30 seconds exposures.
The lens used is the mighty lighty FE T* Sonnar 2,8/35mm ZA reviewed previously with much exhilaration.
The first target was a book in a very dark room only lit by a small stove fire and totally unreadable at one meter. Note these 2 files are FULL SIZE (large !) so you can inspect the results for yourselves.
Sony A7r. 30 seconds @ISO 100, no noise reduction.
Sony A7r. 30 seconds @ISO 100, in camera noise reduction.
What I see is no significant difference at all, and absolutely no shake …
Noise ? What noise ?
Shake ? What shake ? In fact, Jim at The Last Word, performed more scientific tests and found the shutter vibration to be 5 times less than the mirror slap from the Nikon D800.
Note how shallow depth of field is in this configuration (close range, f/2.8, book slightly at an angle) only a top-left bottom-right diagonal showing full resolution. Also note the purple in the background bokeh is due to a purple lamp reflector, not chromatic aberration 😉
For this second test, I pushed exposure by 1.7 stops in camera then added +65 contrast and +40 whites in LightRoom. Both pictures shot at 3200 and displayed here at -2,2 stops for aesthetic reasons.
Sony A7r. 20 seconds @ISO 3200, no noise reduction.
Sony A7r. 20 seconds @ISO 3200, in camera noise reduction.
Below are 2 enlargements of the central region without the 2,2 stop darkening (as shot). But you need no enlargement to notice the effects of noise reduction :
Strong purple fringes appear around bright stars. They look more like sensor blooming than chromatic aberration to me but I have no certainty of this. At any rate, NR ges rid of them and might keep the background a tad cleaner (sorry about the so-so focus, it was dark 😉 )
Largely but not totally unrelated to theoriginal request, here is a crop of the top right corner of the frame to show how well coma is controlled in this lens’s design. And the answer is : exellently.
Very little coma. Click for 100%
#558. The Monday Post (13 Feb 2017) – The anti-anti-Sony rant
# 391. Sony’s A7RII (formerly Leica it or not)
#287. The Susans’ day out.
#217. Should you buy a Pentax 645Z (implied: or a Sony A7R)?
#210. The Leica Macro-Elmarit-R 60/2.8, the soulful engineer
#191. Leica Summicron-R 50/2 : the perfect 50 for the Sony A7r ?
Session expired
Please log in again. The login page will open in a new tab. After logging in you can close it and return to this page.
Pascal, I’d be interested to see the difference in noise between the in-camera noise reduction and no in camera noise reduction, but with reduction performed in Lightroom.
Also, there appears to be a lot of the bullseye colour fringing on both the night shots.
I think that’s just vignetting. Bullseye is mainly visible if you crank up the saturation quite a bit. But I have seen it on occasion, unfortunately. See my post on the “spotty flaw” : http://dearsusans.wordpress.com/2013/12/11/spotty-flaw-in-zeiss-fe-35-28-za-sony-a7r-combo/
Tim, I’ll update the article with this info when I have downloaded the RAW converter this week-end. Do take a look at blog.kassson.com for much more scientific analysis of noise and vibrations if you need numbers. I’m not equiped (or inclined) to perform that sort of test 😉 More soon, though.
I would love to see a test of the multi shot noise reduction technique which stacks multiple images at lower iso , have you had a chance to try this technique in your NR testing?
Hi Chuck, no I haven’t but it shouldn’t be too difficult in Photoshop. I’ll try over the week-end and update. Cheers,
Pascal
Thanks Pascal,
I dont think you have to do a special photoshop , it’s actually a feature in the sony camera that automates the stacking process, instead of a shot at 2000 iso for example, the camera will take 4 shots at iso 500 and stack the exposures for a better noise performance, this is what I was hoping to see tested if possible?
Oh, OK. I didn’t even know about that feature. Let me look for it an report. In the mean time, bwana (other comment) has provided a few samples of the out of camera stacking method. Pascal
From whats being said over on FM, it seems that the vibration is emphasised if you use a lens tripod mount on a longer lens so the camera is “suspended” without any dampening from being bolted down itself.
Tim, Azure, here is the most scirntific shutter vibration measurement I’ve found on the web : http://blog.kasson.com/?p=3757
The reviewer used a 400mm lens and detected a 4-5 pixel shake and regrets that nothing can be done to mitigate this. It’s safe to say lenses below 80 would show none of this blurring whatsoever, but if long lenses are a must for you, this camera may not be the best for you. In my early tests with a 135mm, I’ve never found shutter vibration to be an issue on a tripod, but hand holding the camera can be a difficult task in low light. Hope this helps.
I think your analysis is pretty much on the money. I’ve been testing the A7R for usage as an astro-imaging / nightscape camera on everything from 8mm to 2000mm lens / telescopes.
Noise, even with no in-camera noise reduction, is the best I’ve seen from any camera, even up to ISO 25600. I’d say the A7R at ISO 6400 is about the same as my Canon 60D at ISO 1200-1600. Of course at ISO 25600 I’m stacking several images to improve the signal-to-noise ratio, but the result is quite acceptable! I normally turn all in-camera noise (both long exposure and high ISO) reduction off and handle this in postprocessing. I just like to have complete control over what is kept and what is scrapped BUT the A7R has reduced the amount of work I have to do to get to a final image I’m happy with…
As for the reported (and concern about) shutter vibration, I haven’t seen anything detectable in the 1st 800+ shots I’ve taken but I’m using either a good heavy tripod or telescope mount.
Great info, thanks. Would you like to share some of your stacked images and process (either as a link to your blog or gallery or as a guest post) ? Other readers are interested in this stacking methodology. You’re welcome either way 🙂
The “stacking methodology” is a pretty standard process in astrophotography. There are a number of software packages used. Deep Sky Stacker is freeware. ImagesPlus and PixInsight (both not free) are the two that I use most often. ImagesPlus works very well when stars are well differentiated, i.e.: lots of room between them. PixInsight will stack almost anything with good success, i.e.: my most common usage of PixInsight is nightscapes of the Milky Way where there are literally millions of stars to align.
A couple of recent images can be found at:
http://tinyurl.com/lclsa3l
http://tinyurl.com/mvob79c
http://tinyurl.com/p8f47m3
http://tinyurl.com/pnajr75
http://tinyurl.com/oec656h
http://tinyurl.com/mbs8rtn
Details in captions to images.
Wonderful, thanks a bunch !
[…] with long exposures. Also vibrations and a sidenote on coma (largely absent) in the Sonnar Lens. http://dearsusans.wordpress.com/2013/12/12/186-low-light-long-exposures-with-the-sony-a7r/” And: “there is a reflection problem with the A7r / Sonnar 35/2.8 ZA that produces […]
[…] Low-light long exposures with the Sony A7r […]
The D800e suffers badly from LE noise – loads of tiny white dots across the frame that can only effectively be removed with in-camera NR. It becomes a real pain on 1+ minute exposures. Details of that issue can be found here: http://scottreither.com/blogwp/2012/07/01/nikon-d800-e-long-exposure-issues-problems-2/ . I would like to know if the A7R suffers from the same issue, or if you can safely achieve 2+ minute exposures without the need for in-camera NR? Best way to test this is a 2 min (or preferably longer) exposure of a blue sky at 100 iso.
Hi Dave, I haven’t tested this yet but will do so this week-end and report here. Cheers
Thanks and much appreciated. If you can post full size files from the test that would be great. I’m seriously considering the A7R, but need to be sure first that it doesn’t suffer from the same white dot issue.
All cameras, CCD’s, DSLR’s and mirrorless, will have some hot/dead pixels and these become more prominent with increased temperature, ISO and exposure. Removing them using any type of in-camera noise removal is not a good idea, particularly for astro images, since the camera doesn’t know the difference between noise and faint features/detail in the image. With any of the cameras I use, manual cleanup is normally the best approach.
I’ve been using an A7R for astro-imaging for about two months and find it performs very well, so well that I’ve bit the bullet and have sent the camera out for a full spectrum modification for both IR and astrophotography usage. Since I have yet to lay my hands on an intervalometer for the A7R (one is on order) I don’t have a large number of long exposure images; however, I have pushed a couple of exposures to 5 minutes with reasonable results.
Up to this point my main astrophotography camera has been a full spectrum modded Canon 60D which I have found turns out very clean images. The A7R at ISO6400 gives me about the same image quality as the Canon 60D at ISO1200-1600, which I can live with. With an intervalometer I would reduce the A7R’s ISO and go for longer exposures with, I anticipate, excellent results. As is my normal approach, I will be aligning and stacking probably 25-30 images to get a final result.
I suspect the A7R will come very close to matching the performance of several of my older cooled OSC’s (one shot color) CCD’s… Time will tell?
Actually in-camera LE NR is the best way to remove this issue on the D800e. On my Canon 5D Mk2 before that I never had this problem, just one or two hot pixels that were easily spot removed. But with the Nikon they’re everywhere and the only clean way to remove them without softening the image or spending a lot of time selectively masking with Dust and Scratches is to use the camera’s LE NR function, which shoots a black frame and effectively rids the entire image of them in one go. The downside, of course, is that you have to stand around in the field waiting for it to finish.
To DaveB:
The “standing around in the field” is the main reason I don’t use the LE NR feature; however, it does work quite well.
If I really want to remove static (consistent) noise from subs, I would generate and apply a master dark to the subs BUT maintaining a library of darks is a pain. So I dither the subs, align them and combine them using a Max/Min Exclusion algorithm which nicely eliminates any “spots” not part of the image. I’m not too sure if this would work for anything but astro-imaging but it works well for this application…
I am curently building a 600mm f/5 newtonian scope for deep sky photography and the A7r is part of the testing process until I get my hands on a better CCD. An astrophoto thread would be very welcome if either or both of you are keen to contribute 🙂
Cheers,
Pascal
@bwana – the problem is that it isn’t consistent, and the dark frame subtraction method only really works with frames of the same exposure length. Even then they have to be taken close to the time of the original capture otherwise it’s likely to vary, leading back to manual removal in s/w. A few of us have been batting this problem around with the D800 for a good while, and the only conclusive way to remove the issue is to use in-camera LE NR, which cures it completely. So it’s either extra time in the field that leads to a clean and clear capture, or lengthier time behind a computer that will always end up softening a part of the image. Usually that’s the sky though, where the problem is obvious, so that’s not such an issue.
Having resigned to no easy alternative other than in-camera LE NR with the D800, I’m now looking for alternative cameras that offer similar resolution, in the (perhaps vain) hope that other manufacturers may have resolved the issue or lessened it to such a degree that in-camera LE NR isn’t a necessity. Hope springs eternal, but we’ll see when Pascal has completed tests.
Hi that’s amazing ! Do you display your pictures anywhere ? I’m sure many readers would love to see them. Would you like to write an article on the process here ?
I stumbled across this blog just today. I’ve recently purchased an A7 and I also use it on a telescope for astrophotography. I’ve only done a quick test (some 3 minute exposures, stacked) but the ISO is definitely way better than my old A55.
Here’s the first two images I’ve come up with:- first, the Christmas Tree nebula. The first image (http://www.flickr.com/photos/jmbillings/12882870764/) shows the result of stacking the raw frames in DSS. I’d also tried using a flat frame but this seems to have created a positive vignette rather than removing the dark one. There’s also a “glow” going horizontally across the frame. I’m not sure if this is the camera sensor, a light leak, or a result of the flat frame addition (I’ve not tried without, yet). Cropping and tweaking in Lightroom gives the final result: http://www.flickr.com/photos/jmbillings/12882867854/ – still a bit noisy, more exposures needed.
The second was M44, Beehive Cluster, and again there’s a before (http://www.flickr.com/photos/jmbillings/12882550723/) and after (http://www.flickr.com/photos/jmbillings/12882873654/)
Dear James, these are excellent. Thanks a lot for sharing. I’m a bit puzzled by the gradient as well … It’s almost like the flat frame has been removed twice.
What scope are you using ? It offers quite a broad field of view so I’d expect a refractor but the sexy spikes indicate otherwise.
Lovely pictures. I’d love to see more.
Pascal
The telescope is a 200P/DS – it’s an 8 in reflector, hence the spikes 🙂 This is my astro set on Flickr- http://www.flickr.com/photos/jmbillings/sets/72157625680541035/ although all the earlier shots are with my old A55. Newer stuff I add will be with the A7.
Thanks James. Superb pictures.
@DaveB
“the problem is that it isn’t consistent, and the dark frame subtraction method only really works with frames of the same exposure length” Totally agree and not only for same exposure length BUT ISO and ambient temperature. That is why maintaining a darks library is a pain!
“only conclusive way to remove the issue is to use in-camera LE NR” However, this doubles the time in the field and wastes half the productive imaging time. A 24x10min (4 hr) session turns into an 8 hr session which is simply not practical (and in most cases not doable); the main reason I don’t employ LE NR. In fact with the Canon 60D I don’t use darks either. As I mentioned earlier, dithering works very well in combination with a max/min exclusion combine/integration.
Having looked at the bright pixel problem in the D800/D800E I trust Sony has not imported this problem to the A7R (along with the sensor). Testing to this point would indicate the inherent noise in A7R images is fairly minimal… Regardless, I won’t be using LE NR on the A7R!
Pascal,
Covering a number of your comments in one post:
“Strong purple fringes appear around bright stars. They look more like sensor blooming than chromatic aberration to me but I have no certainty of this. At any rate, NR gets rid of them and might keep the background a tad cleaner (sorry about the so-so focus, it was dark).”
This could be chromatic aberration but I might suggest is could also be the result of focus being slightly off? Not too sure how you obtained focus. I normally use Bahtinov masks on everything; however, I’m finding peak focusing on the A7R works reasonably well, even on stars.
“I am curently building a 600mm f/5 newtonian scope for deep sky photography and the A7r is part of the testing process until I get my hands on a better CCD. An astrophoto thread would be very welcome if either or both of you are keen to contribute.”
A 3000mm Newt, wow! What mount are going to put this monster on? I had a 2700mm scope at one point; however, I almost needed a crane to put it on the mount. Too heavy to handle and far too much focal length for the targets I normally go after… I find my sweet spot for imaging is in the range of 1200-1600mm and as fast a scope (f/5 or better) as I can afford.
As for contributing to an astrophoto thread, sure. However, remember sometimes my contributions could be considered devil’s advocate contributions. I have been known to be a bit unorthodox WRT to astro-imaging :). I do try things that “more experienced” astro-imagers simply shake their collective heads at (and I’ve been at this for about a decade to this point).
Working out of the Misty Valley Ranch observatory:
https://picasaweb.google.com/lh/photo/tqwwOdgIfWaV01SmI3ZtKtMTjNZETYmyPJy0liipFm0?feat=directlink
Hi again, that observatory makes me very jealous. We don’t have such dark skies here.
My newt is finished but the observatory isn’t quite. The mount is a 10 micron GM4000, which I hope will be strong enough. The scope tube is carbon fiber so total weight is relatively low, but the tube is looooong so it will need to be perfectly balanced.
Not sure about the chromatic aberration. You might be right and focus was manual (focus bracketing). But I have seen strong purple flares on many other pictures, including some of that singer in Steve Huff’s Nashville report (I think with the OTUS 55).
You are particularly welcome to contribute if you break the rules 😉 😉 More experienced folk is sometimes right, sometimes used to other gear or techniques that produce technically superior results but not necessarily more interesting. Don’t hesitate if you fancy giving it a try. Astrophotos are so beautiful !
Cheers,
Pascal
Hi – Glad I found this as I have just brought a Sony A7 and I want to find out as much as I can about it by people who use it… the fact you’re reporting on the A7r is good enough, I gather there’s not much difference between the two, save for the pixels?
The issue I have been having with low light long exposure photography, on my A7, is that the live view stops working… e.g. as I change the shutter speed, the live view will get lighter/darker to show how the final image will look. However, there get’s a point as I am making the shutter speed slower that it just stops, so 1/20 fine, 1/15 fine 0.7″ fine… but then it will stop and every stop after 0.7″ you will just get the 0.7″ preview…. so I could have a 30 second shutter speed but it’s showing me the preview for 0.7seconds because somewhere along the line it just gave up!
Are you getting this issue with the A7r??
Thanks
Hi Holly,
I think the A7 is very slightly better than the A7r in terms of noise, but there isn’t much in it. As for the issue you describe, it’s unfortunately the same with the A7r. This is one area where the Olympus EM-1 is vastly superior, thanks to its live refresh of preview during long exposures. Maybe Sony can offer this improvement via firmware update ? 😉
Hello! I just want to offer you a big thumbs up for your great information you have got here on this post. I’ll be coming back to your web site for more soon.
Thanks Dale. I look forward to your comments.