The effects of Sony's star eater

Started Dec 20, 2015 | Discussions
sharkmelley
sharkmelley Senior Member • Posts: 2,710
The effects of Sony's star eater
1

I've been asked many times for a side by side example of the effects of Sony's star eater.

What I've done here is to take a non-bulb image shot with the Sony A7S on a Tak Epsilon telescope (focal length 500mm). I've then applied the bulb-mod spatial filtering to the RAW to provide a direct data comparison of the effects.

Both have been blown up by a factor of 2 to allow easy pixel peeping.

Here's the original:

Here's the simulated bulb-mode version:

It's interesting to see that many stars have had the bright central peak punched out and how this affects the colour balance of many stars. The faintest stars become very much more difficult to see.

The effects are quite striking when you "blink" between one and the other.

Here's an example of a genuine bulb-mode exposure taken with the same equipment:

Trollmanx has posted some excellent comparisons here:

http://www.dpreview.com/forums/post/55841466

Mark

 sharkmelley's gear list:sharkmelley's gear list
Sony a7S Nikon Z6 +1 more
Tristimulus Veteran Member • Posts: 9,787
Re: The effects of Sony's star eater

The last image show excactly what I see when using the A7 with the CZ 1.8/55mm lens at f/4 (have not tried other apertures on the sky). JPGs are worse than the RAW files (more added artefacts but the stars look almost the same). Did even try different RAW converters. Have posted examples here earlier.

When using cheap lenses the stars are bloated and fine detail is washed out because low resolution at low spatial frequencies. Then the star eater is less damaging (to be expected as the images of the stars are bigger and the spatial filtering is size dependent). But low quality optics do not record as faint stars as high quality optics.

So the images shown look very familiar.

This is the reason for me to get some Canon cameras for deep sky work (have skimped on telescope size and the mount to leave some money left for the Canon cameras - kind of balancing act). Still Sony only for me and all my other work. I simply choose to leave the night sky to other cameras handling that single field better (but not without some quirks).

And I am absolutely NOT saying that Sony cameras can not be used for astrophotography.

W5JCK
W5JCK Veteran Member • Posts: 4,004
Re: The effects of Sony's star eater

First, why double the image size turning it into a fuzzy mess? All we have to do in this forum is click an image and it loads it into a separate screen where we can view it full size.

Second, for some reason I've never encountered this so-called star eater issue. Never. With either of my Sony cameras. But then I use good optics and I know how to use the equipment, so perhaps that makes a difference.

-- hide signature --

Jack Swinden
W5JCK, amateur radio operator
An astrophotography hobbyist and amateur radio instructor and examiner. Sony a7 and Sony a6000. https://www.flickr.com/photos/jackswinden/albums

 W5JCK's gear list:W5JCK's gear list
Sony RX100 III Sony a7 Sony a6000 Canon EF 400mm f/5.6L USM Sony E 50mm F1.8 OSS +11 more
Tristimulus Veteran Member • Posts: 9,787
Re: The effects of Sony's star eater

W5JCK wrote:

First, why double the image size turning it into a fuzzy mess? All we have to do in this forum is click an image and it loads it into a separate screen where we can view it full size.

Second, for some reason I've never encountered this so-called star eater issue. Never. With either of my Sony cameras. But then I use good optics and I know how to use the equipment, so perhaps that makes a difference.

That might well be - but I do not think the others here are completly lost in the dark. The Sony star eater effect is there but how it affects diffenent users obviously vary.

Have been into photography (film days and done everything from casette cameras to view cameras and darkroom and lab work) and astronomy (first astrophotos the first year as a teeenager) so I guess I can handle equipment and also evaluate results. I guess! 

-- hide signature --

Jack Swinden
W5JCK, amateur radio operator
An astrophotography hobbyist and amateur radio instructor and examiner. Sony a7 and Sony a6000. https://www.flickr.com/photos/jackswinden/albums

sharkmelley
OP sharkmelley Senior Member • Posts: 2,710
Re: The effects of Sony's star eater
1

W5JCK wrote:

First, why double the image size turning it into a fuzzy mess? All we have to do in this forum is click an image and it loads it into a separate screen where we can view it full size.

The image size was doubled to make the effect clearer to the casual reader without having to zoom in. The fuzzy mess is caused by the Sony bulb-mode algorithm itself, not by the careful size doubling I performed.

Here is a link to an actual Sony A7S bulb-mode raw - 90sec at ISO 2000:

https://drive.google.com/file/d/0B3Ky5pyZvsINaDBraUhvQUVadkk/view?usp=sharing

It will allow you to use your favourite raw converter and display the results however you wish.  Feel free to post any image derived from this data.

Second, for some reason I've never encountered this so-called star eater issue. Never. With either of my Sony cameras.

Fair enough. If it's not causing you any problems then certainly don't go looking for it! Maybe you don't use bulb mode for astrophotography? The effect only happens with bulb mode.

But then I use good optics and I know how to use the equipment, so perhaps that makes a difference.

I find that to be a very strange comment. Take a look at the spot diagrams for Tak Epsilon scope here:

http://www.takahashi-europe.com/en/epsilon-180ED.optics.php

In the centre of the field of view the spot size is around 6 microns i.e. smaller than a pixel of the A7S. My focusing was accurate to just a few microns.

Mark

 sharkmelley's gear list:sharkmelley's gear list
Sony a7S Nikon Z6 +1 more
AstroDan Regular Member • Posts: 275
Re: The effects of Sony's star eater
1

W5JCK wrote:

First, why double the image size turning it into a fuzzy mess? All we have to do in this forum is click an image and it loads it into a separate screen where we can view it full size.

Second, for some reason I've never encountered this so-called star eater issue. Never. With either of my Sony cameras. But then I use good optics and I know how to use the equipment, so perhaps that makes a difference.

First off, Mark thank you for posting these!

Jack, image scale matters quite a bit with this type of filtering. You could put an a7s on say a 3000mm fl scope and see a better result due to the correct/oversampling (seeing dependent). It becomes increasingly destructive as the relative star size shrinks, as they are just rejected as noise.

On a personal level I have found even correctly sampled bulb images from some of these newer chips to have a 'waxy' look to them, but I haven't decided if it is the filtering, or processing approach. I do know that Sony users suggest dithering as a potential solution to this issue. I tend towards increasing sampling instead, as moving around a bunch of stars that have holes punched in them doesn't seem to be much of a solution, but until I give it a try it's hard to comment on it.

Dan

swimswithtrout Veteran Member • Posts: 4,147
Re: The effects of Sony's star eater

W5JCK wrote:

Second, for some reason I've never encountered this so-called star eater issue. Never. With either of my Sony cameras. But then I use good optics and I know how to use the equipment, so perhaps that makes a difference.

"Good optics" has nothing to do with it.  A Zeiss vs. a Vivitar lens makes no difference. It's a noise reduction program written into the onboard camera image processor, that ONLY turns on, when using Bulb in manual mode.

RustierOne
RustierOne Veteran Member • Posts: 4,401
Re: The effects of Sony's star eater

sharkmelley wrote:

I've been asked many times for a side by side example of the effects of Sony's star eater.

Mark, of course you know I am one who has asked for a comparison of the star-eater effect. But what I have been trying to see is a full sensor comparison of the effect. I'm perfectly capable of doing my own pixel-peeping. And I don't doubt at all that the effect is there and very damaging to the image. What I'd like to see is how the entire view is affected. For that I'm asking for something like the following:

  • 30-second non-bulb mode at for instance f/2.8. Display the entire captured frame.
  • 60-second bulb mode at f/5.6. Display the entire frame without cropping - let me do whatever pixel-peeping I want.

In each case the total light is the same. I know that for your Takahashi astrograph it would be difficult to do the above since it doesn't have an adjustable aperture. But perhaps someone with a camera lens can provide the comparison.

It's interesting to see that many stars have had the bright central peak punched out and how this affects the colour balance of many stars. The faintest stars become very much more difficult to see.

The effects are quite striking when you "blink" between one and the other.

Here's an example of a genuine bulb-mode exposure taken with the same equipment:

That effect is certainly very damaging to the image. It looks similar to being out of focus. Whatever prompted Sony to produce such an effect must have had some positive effect for daytime photography. These sensors are highly rated for such use, but miss the mark for astrophotography in bulb mode.

Maybe I'm just displaying my ignorance, but is there some possibility of stacking 30-second (non-bulb) images in an additive rather than average method? It seems like this would be similar to a longer exposure. What would be the effect on noise? But what a lot of hassle to compensate for bad data handling by Sony.

My reason for asking to see full sensor examples of the star-eater effect has to do with my take on pixel-peeped errors in general. If one doesn't like what you see when pixel-peeping, then don't pixel-peep - thus my desire to see the full image. But on the other hand I realize that experts like yourself are able to produce excellent images by getting the best effects visible at the pixel level.

-- hide signature --

Best Regards,
Russ

 RustierOne's gear list:RustierOne's gear list
Sony Alpha NEX-5N Sony Alpha NEX-7 Sigma 19mm F2.8 EX DN Samyang 8mm F2.8 UMC Fisheye Sony E 35mm F1.8 OSS +4 more
sharkmelley
OP sharkmelley Senior Member • Posts: 2,710
Re: The effects of Sony's star eater

RustierOne wrote:

sharkmelley wrote:

I've been asked many times for a side by side example of the effects of Sony's star eater.

Mark, of course you know I am one who has asked for a comparison of the star-eater effect. But what I have been trying to see is a full sensor comparison of the effect. I'm perfectly capable of doing my own pixel-peeping. And I don't doubt at all that the effect is there and very damaging to the image. What I'd like to see is how the entire view is affected. For that I'm asking for something like the following:

  • 30-second non-bulb mode at for instance f/2.8. Display the entire captured frame.
  • 60-second bulb mode at f/5.6. Display the entire frame without cropping - let me do whatever pixel-peeping I want.

In each case the total light is the same. I know that for your Takahashi astrograph it would be difficult to do the above since it doesn't have an adjustable aperture. But perhaps someone with a camera lens can provide the comparison.

The best way to do that would be to compare a 30sec bulb mode image with a 30 sec non-bulb mode image. I don't actually have such a comparison and given the UK weather at present (one imaging night in the last 3 months) I may not have one for some time.

However, I have the next best thing.

Here is a link I posted for Jack, earlier in the thread, to an actual Sony A7S bulb-mode raw - 90sec at ISO 2000:

https://drive.google.com/file/d/0B3Ky5pyZvsINaDBraUhvQUVadkk/view?usp=sharing

Here is a link to Sony A7S non-bulb raw of the same subject taken a few weeks later - 30sec at ISO 2000:

https://drive.google.com/file/d/0B3Ky5pyZvsINQ1NzSlY3Tl90Z2c/view?usp=sharing

The exposure times are different but that has no effect on the spatial aliasing.

I realise it is not like for like in terms of exposure but it will allow a comparison of a full sensor image if you compensate in post-processing. As I said to Jack, feel free to post any image derived from this data, in the context of this discussion or similar discussions.

Regards,

Mark

 sharkmelley's gear list:sharkmelley's gear list
Sony a7S Nikon Z6 +1 more
RustierOne
RustierOne Veteran Member • Posts: 4,401
Re: The effects of Sony's star eater

sharkmelley wrote:

RustierOne wrote:

sharkmelley wrote:

I've been asked many times for a side by side example of the effects of Sony's star eater.

Mark, of course you know I am one who has asked for a comparison of the star-eater effect. But what I have been trying to see is a full sensor comparison of the effect. I'm perfectly capable of doing my own pixel-peeping. And I don't doubt at all that the effect is there and very damaging to the image. What I'd like to see is how the entire view is affected. For that I'm asking for something like the following:

  • 30-second non-bulb mode at for instance f/2.8. Display the entire captured frame.
  • 60-second bulb mode at f/5.6. Display the entire frame without cropping - let me do whatever pixel-peeping I want.

In each case the total light is the same. I know that for your Takahashi astrograph it would be difficult to do the above since it doesn't have an adjustable aperture. But perhaps someone with a camera lens can provide the comparison.

The best way to do that would be to compare a 30sec bulb mode image with a 30 sec non-bulb mode image. I don't actually have such a comparison and given the UK weather at present (one imaging night in the last 3 months) I may not have one for some time.

However, I have the next best thing.

Here is a link I posted for Jack, earlier in the thread, to an actual Sony A7S bulb-mode raw - 90sec at ISO 2000:

https://drive.google.com/file/d/0B3Ky5pyZvsINaDBraUhvQUVadkk/view?usp=sharing

Here is a link to Sony A7S non-bulb raw of the same subject taken a few weeks later - 30sec at ISO 2000:

https://drive.google.com/file/d/0B3Ky5pyZvsINQ1NzSlY3Tl90Z2c/view?usp=sharing

The exposure times are different but that has no effect on the spatial aliasing.

I realise it is not like for like in terms of exposure but it will allow a comparison of a full sensor image if you compensate in post-processing. As I said to Jack, feel free to post any image derived from this data, in the context of this discussion or similar discussions.

Thanks, Mark for the full size images to work with. So I imported each one into Lightroom 5.7 and did my best with a quick edit of the RAW files. I couldn't get the colors of the nebula to look the same, but that's not the issue with star-eater. Below are the two images. In one sense it's not fair since the star eater image is exposed 3X as much as the non star-eater.

#1 - Non star-eater image

#2 - Star-eater image, (EXIF is wrong, the RAW indicates 90-seconds exposure)

While I prefer #2 because of its longer exposure bringing out the nebula better, it is missing the faintest stars. These fainter stars are found right in with the noise on image #1. Also the star images are smaller in diameter in #1 (non star-eater). Strangely I saw no evidence of the cores of the stars being punched out by star-eater. But that may be due to how I processed the images.

If it was possible to get a non star-eater image of 90 seconds exposure, I'm sure it would be more to my liking with not only well-exposed nebula but more, smaller sized star images. When viewed closely (not even pixel-peeping) the star-eater image seems to be smoothed, almost a plastic look to it. Noise was not evident, but neither are the very faintest stars. In my processing I did not apply any noise reduction to either image. So the lack of noise with star-eater is due to in-camera processing.

So the overall effect of the star-eater data processing doesn't look bad at all. But just knowing that it could have been better would make me very reluctant to spend that much money on an A7S. Maybe if they fix this problem in later versions it would be different. Just a menu item to switch off this data smearing would give both daytime and nighttime photographers what they want. We astrophotographers are tough on sensors and lenses.

Thanks, Mark for giving me the images to satisfy my curiosity. You are right that blinking between images when zoomed in shows the difference quite clearly.

-- hide signature --

Best Regards,
Russ

 RustierOne's gear list:RustierOne's gear list
Sony Alpha NEX-5N Sony Alpha NEX-7 Sigma 19mm F2.8 EX DN Samyang 8mm F2.8 UMC Fisheye Sony E 35mm F1.8 OSS +4 more
sharkmelley
OP sharkmelley Senior Member • Posts: 2,710
Re: The effects of Sony's star eater

I can see plenty of stars with holes. Just look in the centre of the 90 sec image you posted. Look for medium bright stars - not the saturated stars.

Mark

 sharkmelley's gear list:sharkmelley's gear list
Sony a7S Nikon Z6 +1 more
RickColman Regular Member • Posts: 351
Work Around: The effects of Sony's star eater ???

I have an A7 and A7S, and have become very addicted to camera astrophotograpy, and night landscapes involving sky backgrounds. I even have a few good ones! I sold my Canon gear (6D, 7d are lenses) to fund the Sony purchases. By and large, I have been happy with the move.

I have, however, been lusting after more than a 20+ exposure time and was about to order a small tracking device; however, in light of the "star eater" effect - which looks plainly evident to me - I am wondering whether to even bother with longer exposure times?

SO :

IS THERE AN EASY WORK AROUND ???

Rick Colman

-- hide signature --

long exposure astrophotography with large (24-40") scopes back in the day

-- 40+ years in photography, mechanical engineering and software development

-- {don't you just hate qualifications signatures!}

Tristimulus Veteran Member • Posts: 9,787
Re: Work Around: The effects of Sony's star eater ???

RickColman wrote:

I have an A7 and A7S, and have become very addicted to camera astrophotograpy, and night landscapes involving sky backgrounds. I even have a few good ones! I sold my Canon gear (6D, 7d are lenses) to fund the Sony purchases. By and large, I have been happy with the move.

I have, however, been lusting after more than a 20+ exposure time and was about to order a small tracking device; however, in light of the "star eater" effect - which looks plainly evident to me - I am wondering whether to even bother with longer exposure times?

SO :

IS THERE AN EASY WORK AROUND ???

Rick Colman

My easy work around was getting Canon cameras for astrophotography, but this was probably NOT the answer you were hoping for. 

Still use Sonys for everything but the night sky.

RickColman Regular Member • Posts: 351
Re: Work Around: The effects of Sony's star eater ???

RickColman wrote:

I have an A7 and A7S, and have become very addicted to camera astrophotograpy, and night landscapes involving sky backgrounds. I even have a few good ones! I sold my Canon gear (6D, 7d are lenses) to fund the Sony purchases. By and large, I have been happy with the move.

I have, however, been lusting after more than a 20+ exposure time and was about to order a small tracking device; however, in light of the "star eater" effect - which looks plainly evident to me - I am wondering whether to even bother with longer exposure times?

SO :

IS THERE AN EASY WORK AROUND ???

Rick Colman

My easy work around was getting Canon cameras for astrophotography, but this was probably NOT the answer you were hoping for. 

Still use Sonys for everything but the night sky.

Well, I used a Canon 6d for quite a while, and it was not a great experience. Focusing was difficult, I could not frame a night sky landscape without shining big flashlights on the scene, excessive noise above 3200 iso. Heavy, bulky, etc.

I would like to explore third party software approaches.

Does anyoNE know if I can set a manual exposure time above 30 sec with a 3rd party app?

R.

rnclark Veteran Member • Posts: 3,991
Re: Work Around: The effects of Sony's star eater ???

RickColman wrote:

Well, I used a Canon 6d for quite a while, and it was not a great experience. Focusing was difficult, I could not frame a night sky landscape without shining big flashlights on the scene, excessive noise above 3200 iso. Heavy, bulky, etc.

Perhaps you had a defective camera, as the 6D is one of the lowest noise cameras around.  At ISO 3200 it has read noise of 2.2 electrons, no banding issues.  Focusing is simple in live view: zoom in 10x and examine a star image.

RickColman Regular Member • Posts: 351
Re: Work Around: The effects of Sony's star eater ???

RickColman wrote:

Well, I used a Canon 6d for quite a while, and it was not a great experience. Focusing was difficult, I could not frame a night sky landscape without shining big flashlights on the scene, excessive noise above 3200 iso. Heavy, bulky, etc.

Perhaps you had a defective camera, as the 6D is one of the lowest noise cameras around.  At ISO 3200 it has read noise of 2.2 electrons, no banding issues.  Focusing is simple in live view: zoom in 10x and examine a star image.

I don't think it was defective because there were no other defects in the image, and the amount of apparent visual noise increased in a predictable fashion as ISO was increased.

The 6d is the best of the Canon bodies for low light photography, better than my buddie's 5d mark three for example. But it is not semi-isoless like the sony sensor in the a7s.

Anyways, I should have said very difficult to frame a scene rather than focus ... however, it seems like focusing star images is easier on the sony because the lcd looks brighter to me than the live view image on the canon.

Anyways, I am not knocking the 6d because I used it successfully for more than a year before switching. I also did not mean to hijack the thread.

I am upset about the star eater problem on bulb exposures. If I had known about that problem plus the compressed raw file situation, I might not have switched to sony at all. Grrrr.

RickColman Regular Member • Posts: 351
Re: The effects of Sony's star eater

It occurs to me that using an intervalometer for multiple timed exposures requires setting shutter speed to BULB.

Won't this kick off the dreaded "STAR EATER" problem where Sony decides to process out parts of an image (in this case faint stars and centers of others) !!!

Rick Colman

sharkmelley
OP sharkmelley Senior Member • Posts: 2,710
Re: The effects of Sony's star eater

RickColman wrote:

It occurs to me that using an intervalometer for multiple timed exposures requires setting shutter speed to BULB.

Won't this kick off the dreaded "STAR EATER" problem where Sony decides to process out parts of an image (in this case faint stars and centers of others) !!!

Rick Colman

Yes, using an intervalometer in bulb mode will trigger the dreaded star-eater.  But if your timed exposure is 30sec or less there is no need to set the camera to bulb mode - the intervalometer can still trigger the exposure.

Mark

 sharkmelley's gear list:sharkmelley's gear list
Sony a7S Nikon Z6 +1 more
rafalczak Regular Member • Posts: 102
Re: The effects of Sony's star eater

sharkmelley wrote:

RickColman wrote:

It occurs to me that using an intervalometer for multiple timed exposures requires setting shutter speed to BULB.

Won't this kick off the dreaded "STAR EATER" problem where Sony decides to process out parts of an image (in this case faint stars and centers of others) !!!

Rick Colman

Yes, using an intervalometer in bulb mode will trigger the dreaded star-eater. But if your timed exposure is 30sec or less there is no need to set the camera to bulb mode - the intervalometer can still trigger the exposure.

Mark

As long as you can live with 30 seconds single exposure limitation, you can avoid BULB mode by using manual mode 30s shutter and burst mode triggered and held.

 rafalczak's gear list:rafalczak's gear list
Panasonic Lumix DMC-TS2 Sony RX100 III Fujifilm X-T2 Fujifilm XF 18-55mm F2.8-4 R LM OIS Fujifilm XF 27mm F2.8 +2 more
sharkmelley
OP sharkmelley Senior Member • Posts: 2,710
Re: The effects of Sony's star eater

rafalczak wrote:

sharkmelley wrote:

RickColman wrote:

It occurs to me that using an intervalometer for multiple timed exposures requires setting shutter speed to BULB.

Won't this kick off the dreaded "STAR EATER" problem where Sony decides to process out parts of an image (in this case faint stars and centers of others) !!!

Rick Colman

Yes, using an intervalometer in bulb mode will trigger the dreaded star-eater. But if your timed exposure is 30sec or less there is no need to set the camera to bulb mode - the intervalometer can still trigger the exposure.

Mark

As long as you can live with 30 seconds single exposure limitation, you can avoid BULB mode by using manual mode 30s shutter and burst mode triggered and held.

That's right. Or you can use single shot mode and set the intervalometer to trigger the shutter every 32 seconds, say.

Mark

 sharkmelley's gear list:sharkmelley's gear list
Sony a7S Nikon Z6 +1 more
Keyboard shortcuts:
FForum MMy threads