Star Eater: Documentation of an Issue with the Sony Cameras for Astrophotography (and How to Fix It)

This article includes documentation of the so-called “Star Eater” issue with certain Sony cameras and long exposures of the night sky. 

Update October, 2020:

According to multiple sources, including this video from Alan Stankevitz, the Sony a7SIII exhibits noticeable star eating noise reduction in its video mode at ISOs of 102400 and higher.

Update December, 2018:

With the discovery of a workaround, and improvement in the noise reduction algorithm in the Sony Mark III Alpha cameras (a7III and a7RIII), I have updated the title of this post and I have appended the content of my workaround post and elements of my a7III review to show what users can expect from these cameras. I still recommend Sony cameras for astrophotography, especially the new a7III and I still use the original a7S as my primary astrophotography camera. Those interested or concerned about the issue can read about how to avoid it here.

Documentation of the Problem

In 2016, Sony purportedly made firmware changes to their two flagship a7 series cameras (the a7RII and a7SII). The update (believed to be 3.30 on the a7RII and 2.10 on the a7SII) was supposed to include improvements for radio controlled lighting and overall camera stability and temperature control. But hidden in these improvements is a change that affects the image quality when shooting long exposures, particularly astrophotos. The problem has been dubbed “Star Eater” by others in the astro community. The issue also affected, from day one, the original line of a7 cameras when used in Bulb mode and the problem persists through out Sony’s latest line of cameras including the a9 and a7RIII.

sony-alpha-star-eater-comparison
Full resolution examples of the “Star Eater” issue are available for download here (.zip, 8.4MB). (a7S, 30s camera timed vs 96s Bulb)

The “Star Eater” problem is a form of software spatial filtering designed to reduce noise in photos, particularly hot pixels. Unfortunately, the rather rudimentary filtering algorithm that Sony is using easily mistakes sharp pinpoint stars for noise, nearly deleting them from the image or greatly reducing their brightness. The result is an astrophoto with less stars and the appearance of diminished resolution. Sony a7RII, a7SII and a7RIII cameras exhibit this problem for all exposure times longer than 3.2″. Furthermore, the “Star Eater” issue affects RAW images (whether uncompressed or not) and cannot be disabled by any means. There are no user selectable settings that will prevent these cameras from eating stars.

sony-alpha-star-eater-demo
Comparison of an unaffected exposure with a “Star Eater” exposure at 100% crop. (a7S, 30s camera timed vs 96s Bulb, tracked)

I have confirmed the issue myself and it has been discussed and analyzed many times on other places around the online photography community, particularly on the DPReview forums, Cloudy Nights, Jim Kasson’s Last Word, and more recently on Sony Alpha Rumors.

Bad spatial filtering isn’t new to Sony’s Alpha cameras. Even with the older firmware, all recent Sony alpha mirrorless cameras including the a6000, a6300, a6500, a7S, a7R, a7, a7II, a7SII and a7RII exhibit, in different ways, the “Star Eater” issue when using the Bulb exposure mode (my examples above are from the a7S in Bulb Mode). Untracked landscape astrophotography rarely requires exposures longer than 30″ so I found that the issue rarely affected my photography. (Most of my exposures are made between 5 and 20 seconds.)

The Workaround

Sony a7SII Continuous Drive Mode
Star Eater Workaround: Enable Continuous Drive Shutter Mode

In a December 2017 post, Jim Kasson talked about a potential workaround on the a7RII. The fix is certainly counter-intuitive: enabling continuous drive mode while shooting long exposures. In his analysis, Jim shows that by enabling the continuous drive mode, the a7RII no longer exhibits the same behavior of the star-eating algorithm. Jim’s original analysis was only of the a7RII so I was curious to see if the same workaround would work on the a7SII. I had a friend supply some sample dark-frames from his a7SII and I took a close look at the files. Sure enough, enabling continuous mode seemed to prevent the star eater problem from occurring. I sent the frames to Jim for his final analysis of the a7SII star eater workaround which he has published here.

  • So the good news: a7RII and a7SII users can circumvent the Sony star eater issue by shooting with continuous shutter mode enabled. Either regular “Continuous” or “Continuous Speed Priority” shutter modes will work.
  • The bad news: by enabling continuous shutter mode, the color depth of the raw files is reduced to 12-bit. This means that the file is losing some dynamic range capability and may show some increase in noise. In all practicality, the results should be more desirable for astrophotography than the star-eaten files.

This workaround is a welcome find for owners of the a7RII and a7SII who enjoy shooting astrophotography but it’s still only a partial fix. Using the workaround, bit-depth is slightly reduced, limiting the dynamic range of the resulting photographs.

Should you care?

Maybe. I’ll admit that it is a problem that requires a fair bit of pixel peeping. Fellow landscape astrophotographer Michael Frye has made a great analysis of what you can expect from the a7RII. All told, some photographers might never even notice the issue. But as our community shifts more and more towards the enthusiast, to the photographer who really cares about the finest capability of their equipment, it’s an issue that I want to document.

sony-fe-16mm-f35-sel057fec-astrophotography-review-5

Without using the workaround, the a7SII in particular is affected very greatly by this issue because of its lower resolution sensor. It was a camera that launched with praise about its low light capability and now I highly discourage you use an a7SII for astrophotography. Want to know how each camera is affected by the issue? Here’s a summary of the differences between each camera and recommendations on how to deal with it:

  • Sony a7
    • Camera timed exposures up to 30s unaffected.
    • Bulb exposures affected.
    • Recommendation: Don’t shoot astrophotography with Bulb exposures.
  • Sony a7R
    • Camera timed exposures up to 30s unaffected.
    • Bulb exposures affected.
    • Recommendation: Don’t shoot astrophotography with Bulb exposures.
    • This camera not recommended for tracked astrophotography.
  • Sony a7S
    • Camera timed exposures up to 30s unaffected.
    • Bulb exposures affected.
    • Because of the lower resolution sensor (12MP), star eater issue for Bulb exposures is significantly more apparent.
    • Recommendation: Don’t shoot astrophotography with Bulb exposures.
    • This camera not recommended for tracked astrophotography.
  • Sony a7II
    • Camera timed exposures up to 30s unaffected.
    • Bulb exposures affected.
    • Recommendation: Don’t shoot astrophotography with Bulb exposures.
    • This camera not recommended for tracked astrophotography.
  • Sony a7SII
    • Camera timed exposures longer than 3.2s affected.
    • Bulb exposures affected.
    • Because of the lower resolution sensor (12MP), star eater issue is significantly more apparent.
    • Recommendation: Use continuous shutter mode. Don’t shoot astrophotography with Bulb exposures.
  • Sony a7RII
    • Camera timed exposures longer than 3.2s affected.
    • Bulb exposures affected.
    • Continuous Low and High modes are a potential workaround, but with a reduction to 12-bit color depth.
    • Recommendation: Use continuous shutter mode. The issue is mostly hidden by this camera’s higher resolution sensor. Don’t shoot astrophotography with Bulb exposures.
  • Sony a9
    • All timed exposures affected.
    • Bulb exposures affected.
    • Recommendation: Don’t shoot astrophotography with Bulb exposures. The issue can be partially hidden by this camera’s higher resolution sensor. Use longer than recommended exposure times to increase star trailing to reduce the effects of star eater.
  • Sony a7RIII
    • Camera timed exposures longer than 3.2s affected.
    • Shows improvement over a7RII.
    • Recommendation: The issue is mostly hidden by this camera’s higher resolution sensor. Use of long enough exposure times to increase star trailing will reduce the effects of star eater.
  • Sony a7III
    • Camera timed exposures longer than 3.2s affected.
    • Shows improvement over previous generation
    • Recommendation: The issue is mostly hidden by this camera’s higher resolution sensor and improved algorithm. Use of long enough exposure times to increase star trailing will reduce the effects of star eater.

Improvements in the Sony a7III

The below section is an excerpt from my a7III review.

The a7III exhibits a similar, but different characteristic noise filtering as previous generations. The filtering reduces noise and diminishes the brightness of small dim stars in an image for exposures of 4 seconds or longer. An easy way to see the effects of star-eater on the a7III is to simply compare a 3.2s exposure (left) with a 4s exposure (right):

Sony a7III Star Eater
A comparison between a 3.2s exposure and a 4s exposure makes apparent that the a7III still has some built-in noise reduction that affects stars.

There’s definitely a difference in the 4 s exposure when compared to the 3.2 second exposure. At 4 seconds, noise is reduced dramatically and dim stars are further diminished in brightness, a clear indicator that Sony is filtering out the higher frequency noise and taking some stars with it. So star-eater is still present, but how does the problem compare with previous generations and how detrimental is it to actual shooting?

A 3.2 to 4 second exposure is a little short for most landscape astrophotos. Let’s take a look at a slightly more realistic 8 second exposure, compared with a previous generation camera that exhibits the star-eater issue (the a7S in Bulb-timed shooting mode). At 8 seconds long, on an 18mm lens, the stars should hopefully “burn-in” more and be less susceptible to the star-eater issue.

The comparison below shows an 8s star-eater exposure from the original generation a7S on the left (8s, Bulb-timed) versus a standard camera-timed 8s exposure from the a7III on the right. The a7S image was scaled to match for comparison purposes.

Sony a7S vs a7III Star Eater
Compared to a star-eaten Bulb-timed image on the older Sony a7S, the a7III does a much better job at resolving dim stars.

As you can see, the 8s a7III (right) shot looks a lot better than the star-eaten a7S Bulb shot (left). The a7S Bulb shot (left) shows some weird color shifted or blocky looking stars that appear a little bit defocused as a results of the old a7S’s spatial filtering (star-eater) in Bulb mode. The a7III shot shows more dim stars and stars appear markedly sharper than in the old a7S Bulb shot. Stars look “normal” in the a7III shot. This result is great news for astrophotographers looking to get the a7III.

If we compare details from that same 8s image from the a7III with an 8s non-star-eater (camera-timed) image from the a7S, differences become significantly less noticeable. The a7S image was scaled to match for comparison purposes.

Sony a7S vs a7III Non Star Eater Comparison
In real-world shooting, the a7III output nearly matches that of a non-star eaten image from the older Sony a7S.

Here, both the a7III (left) and a7S (right) seem to be on par with each other in this direct 8S exposure comparison. The frequency and sharpness of stars look pretty much the same between both images. Upon very close inspection, I’d probably give the original a7S a very slight advantage in terms of noise but both cameras seem to do a good job at resolving fine stars.

So, there’s still some noise filtering in the Sony a7III, but in practice, it’s much improved from previous generations. Sony has obviously tweaked their noise algorithm a bit and the result is much more acceptable for typical night photography exposures. The a7III’s higher resolution sensor, when compared to the a7S probably also helps here. While star-eater is improved, it doesn’t hide the fact that the Sony is still doing some funky stuff to the a7III long exposure RAWs.

Camping at Consultation Lake, Mt. Whitney Trail
Camping at Consultation Lake, Mt. Whitney Trail, California. Sony a7III, 18mm @ f/2.8, 20s, ISO 3200.

We wish that Sony did not apply a noise reduction algorithm as such to the a7III’s RAW files. RAW should be raw, unfiltered and un-baked. Ultimately, however, we really love most of the astrophotos that we’ve made from the a7III and we think that they speak for themselves.

References

For reference, here are a number of online sources that discuss the “Star Eater” problem:

DPReview:

https://www.dpreview.com/forums/post/58709160
https://www.dpreview.com/forums/post/55841466

Cloudy Nights:

https://www.cloudynights.com/topic/498339-sony-a7s-star-eater-algorithm/
https://www.cloudynights.com/topic/505754-another-real-world-example-of-sonys-star-eater-problem/
https://www.cloudynights.com/topic/558230-star-eater-in-action-sony-a7rii/

SAR:

http://www.sonyalpharumors.com/specific-a7sii-astrophotography-fix-request/#disqus_thread

Sony Community:

https://community.sony.com/t5/Alpha-NEX-Cameras/Star-eater-in-bulb-mode/td-p/508740

From Jim Kasson:

http://blog.kasson.com/the-last-word/spacial-filtering-of-raw-images-by-sony-a7s-a7ii/
http://blog.kasson.com/the-last-word/sony-a7rii-long-exposure-spatial-filtering-with-fw-3-30/
http://blog.kasson.com/the-last-word/16486/
http://blog.kasson.com/the-last-word/lenr-and-sony-a7rii-fw-3-30-lowpass-filteering/
http://blog.kasson.com/the-last-word/reverse-engineering-the-sony-a7rii-long-exposure-spatial-filtering/
http://blog.kasson.com/the-last-word/sony-a7rii-bulb-spatial-filtering/

Jim Kasson has written about a possible workaround on the a7RII by using “Continuous High or Continuous Low” modes.

275 Replies to “Star Eater: Documentation of an Issue with the Sony Cameras for Astrophotography (and How to Fix It)”

  1. I just received a brand new Sony A7S and I’m dismayed after reading your email, Norman. The software version installed from the factory is the 3.30 version. I am unclear if you are saying the 3.30 version is only affecting the A7S II as far as camera exposures longer than 3.2 secs on bulb, or if this version is also affecting the A7S and my A7R which also has the 3.30 firmware installed?

    Perhaps I could live with only choosing the camera-based time options instead of bulb, but what about work done with an intervalometer that is choosing other shutter time options?

    I am ready to send this camera back to B&H, but will hold off a little until I’m clear of the limits of this new A7S as far as star-eating goes. I take it you never upgraded your firmware on your A7S, right?

    Thanks,
    Marty

    1. Send back ASAP. It also affects the A7S. I’m in the process of dumping all of my sony gear for this. At these prices and investment, this is absolutely unacceptable. Years ago I was bit by Sony and said I would never give them another chance. However, in recent years, they seemed to be producing great equipment. Sadly, their own developers now get in the way and hobble otherwise amazing gear with algorithms that crippled the performance. I personally would send that back as fast as you can and stay far away until these idiots make a change permanently. This is unacceptable and we as consumers need to make them hear us – with our dollars.

    2. Does not affect the original A7s.
      This camera always had this issue in bulb mode, but not in any other.
      Check out my Instagram for a long exposure made with the A7s latest software.

    1. As far as I know, the latest a7S firmware 3.20 does not change the behavior of Star Eater. It DOES affect Bulb still but does not affect timed exposures up to 30s in M and S modes.

  2. Like someone else mentioned before, I think that, although this is a very serious issue, your recommendations still apply, of course within a certain limit now. We are all hoping to see this situation fixed and we all know the sensors they use are probably the best ones in low-light performance and are amazing for landscape astrophotography. This is definitely a set-back for sony and I understand if you would start to consider to go away from sony bodies because of this reason (I myself own an a6000, and have been on the lookout for the next sony FF line (I was hoping for a camera like the a7II, but built like the a7SII and a7RII, with the better moisture and dust resistance, and the improved sensor, which would not be with 12MP, so around 24 MP, but all the goodies of the a7RII and a7SII)).

    Maybe this was not your intention, but as I read your post, towards the end, I had this feeling of outrage and betrayal, and while it’s true that any person that has bought one for this particular purpose, this is a betrayal of our needs, you did not know this before, you took amazing landscapes and astrophotos and this was not an issue for your type of photography (previously). And at the moment you are aware of the issue and you did the right thing in reaching out to all the community you built to try to get even a greater response and movement to make sony acknowledge (and excuse my language now, but I really wanna say it) they f*cked up and that they will work on a fix. However, in your shoes, although this is of course a personal website, you don’t need to bind to those rules, I’d have liked to see a kind of more neutral and diplomatic post. The users here, can rant, show how they hate sony for this, etc. but for me, the image I have of you, as a kind of a leader, in this kind of matter I would just like to see a description of the problem (you did that), that you yourself, like many others are disappointed at the moment (did that), and that we should all team together to voice our disappointment (you did that as well), but not hate the fact that you recommended these cameras before, it was sony’s fault, not yours, and that you don’t want to endorse their bodies (although you specifically stated “…until we see a fix”) or firmware updates (specially if it is going to be through a firmware update that we will get this issue fixed). The feeling I got was more (and again, I guess, from all the posts and videos of you I read and watched that this was not your intention) that you really despised the idea of getting a sony camera with the newest firmware. It’s understandable, but I think somewhere in the lines of “These are still great cameras in low-light, etc. but if you were looking into buying one with astrophotography in mind, please be aware that there is a very serious issue with the way the image is processed and this might be very detrimental to your end results, and there is no foreseeable fix to this issue at the moment. If you need to buy a new body now, with good astrophotography capabilities, refrain from buying a sony body. If this is not an urgent matter, just seat tight for some time and cross your fingers that sony will give us the fix to this problem in a timely manner…”
    This way you show also hope that there will be a fix and that you still keep your faith that in the future these bodies will still kick-ass in astro.
    And although sony really made a big mistake, sometimes these are not intentional, and may be harder to fix than we think. It really is a big mess right now, specially being that this was already included in the BULB mode before and now it extended to such low exposures.

    Like I said, I was looking into sony bodies for the future, specially me liking astro so much, also wanting to do some tracked exposures, but with objectives (wide-field), not with telescopes, and if this issue is not fixed until that time comes, I will have to go for something else…which is a pity because i have been investing in lenses and gear specifically thinking that I would not move from sony.

    Sorry for the long post, with a rant about a rant, I will also send them an email, with all my thoughts and views.

    In the meantime, as this is my first post here, thank you so much for all the time and passion you put into this. It’s a great website, and great youtube videos.

    All the best and good luck, we all have lots of hope on you!

    1. Thanks José. This post is certainly one of disappointment but also one of hope and a call to action for Sony. I haven’t sold off all my Sony equipment yet or told anyone else to do so. I merely want to prompt a swift response and fix by Sony.

  3. A couple of questions based on this article:

    1. I have a Sony a6000 and a6300 largely because of the amazing work here at this site and the reviews about those two models. I used the a6000 (modified by Spencer’s Camera) for astrophotography work (bulb mode – i.e. longer than 30 second exposures). Have I wasted my money and time? Do I need to just completely “cut bait” on the Sony route and get a modified canon dslr for long-exposure astrophoto work with my telescope?

    2. I use the Sony a6300 for wide-field work with a Rokinon 12mm and 24mm lens. These exposures stay under the 30 sec. mark. Am I losing quality with this camera since most of my subs are longer than 3.2 sec.?

    Thanks. This can begin to get confusing so I appreciate any help since my budget is limited and this spring and summer I don’t want to be wasting more time and money in this if Sony such crapped the wagon with stupid firmware decisions.

    1. As far as I know, the a6000 and a6300 are affected in Bulb mode only. For wide-field work, you should be OK up to 30s exposures.

  4. I find this very sad because I was seriously thinking about buying a soney A6000 for a compact SLR that was great at low light and astrophotography but now I am not sure. Are the cameras still good at taking other types of pictures they are capable of? Are they still good at low light and night photography? or is it only just bad for stars now?

    1. I use an a6000 running firmware 3.20.

      I’m not much of an astrophotographer, but the a6000 seems to only be affected when in BULB mode. Outside of it, it seems to not be affected.

      General low-light photography is unaffected. The NR only affects exposures longer than 4 seconds in the case of the A7RII and A7SII at least.

  5. Hi Ian,
    Does this issue affect all exposure times greater than 3.2s on the a7ii, or does it only affect bulb mode?
    I use a7ii with firmware version 3.30 for astrophotography, and would really like to know the answer to this question. Thanks!

    1. I shot a black background to experiment with the noise reduction system yesterday, there does seem to be a reduction in noise when the exposure is greater than 3.2s. I managed to turn this off in the settings(Long Exposure NR and High ISO NR in tab 6 of general photography settings), and this does seem to really increase the level of noise.

    2. I am actually not sure the exact behavior of the a7II and whether it suffers the same issue as the a7SII or a7RII when using exposures longer than 3.2″. I do, however, know that the a7II is affected when using Bulb exposures.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.