Reading mode: Light Dark

Long exposure noise II - surprisingly large noise jump at 1600

Go to last post
D

Dominique Dierick

Veteran Member
Messages
4,495
Reaction score
53
Location
BE
While doing noise tests for long exposure astrophotography with D800, a strange jump in noise increase is seen going from ISO 800 to ISO 1600. Someone may be able to explain this from a technological point of view. I find it quite dramatic.

Exposure time was 300s, no NR at all.

The first two images are the first of a series of 12 at each ISO (800 and 1600).

The last two images are also the last images of the series of 12. Not unexpected, noise is substantially higher due to the sensor warming up over the elapsed time. Some camera's, like Canons 450D are good in keeping temperature steady. From what I know from other camera tested, D800 seems average in behavior.

Mind, noise is still ok, these samples are extremely stretched. So, does it mean you cannot make images of the nightsky with it? No, but you need to do some more pp here, and preferably consider to shoot ISO 800, not higher unless you have too or images do not need to get stretched in levels a lot.

First image ISO 800





First image ISO 1600





Last image ISO 800





Last image ISO 1600





--
My equipment: Nikon D3s, D800, EOS 450D
The ladies equipment: Panasonic LX5 and GH2 for video
 
My god, It's full of stars!
 
Why such a high ISO? The longer the exposure you're planning, the lower the ISO should be set. Stars/star trails/star fields should be shot at the camera's native ISO (ISO100 in this case). High ISO equals more noise in long exposures. The high ISO/low noise capabilities of the D3, D700, D4 and D800 are expressly designed to make relatively low noise captures at high ISO and using relatively fast shutter speeds (anything faster than 1/15s I think).

--
Howard Carson
Managing Editor, Kickstartnews.com
 
Interesting..... However I know that if the sensor on these cameras get noisy when the ambient temperature is in the 70's & 80's at night. I always shoot in the winter when it's colder. You need to open those up in Nikon's newest version of Nikon capture NX2, it has a setting called "astro noise reduction" and it will get rid of all of that noise. Here is a link showing some of the same noise that you have in your samples:

http://forums.dpreview.com/forums/readflat.asp?forum=1021&message=40152389&q=astro+noise&qf=m

NX2 astro noise remover tool will / should get rid of that. Was the camera cold from night air or was it fairly warm out?

Todd
--
http://www.toddhargisphotography.com

http://toddhargisphotography.com/Weddings

http://www.pbase.com/todd991

 
This is a stress test to see what happens when it gets pushed.
 
Posted this in the Canon conference as a reply to someone asking the same question:

There are some reasons to use higher ISO for anything else but a casual astrophoto of startrails.

The tracking mechanism of equatorial mounts, needed to compensate for earth rotation is not always perfect, so many people prefer the shorter exposures of higher ISO to avoid ruining longer exposures at lower ISO by mechanical shortcomings.

Another reason is getting rid of random noise by averaging out multiple images. Random noise can be residual noise not filtered out by darkframes (in a camera without temperature control, darkframes will not always match exactly), cosmic rays or just man made satellites and airplanes. Suppose you would use a single one hour exposure and a satellite moves through the field of view. Your effort would be zapped, or you would need to touch up the image to get rid of the trail.

If you make 12 x 5 minutes, combining image with sigma reject algorithms will statistically remove the trail nearly perfectly.

Now, some objects are also very faint, with ISO 100 it would just take too long too.

Now, why exactly ISO 1600? That's not mandatory, one could use 800 or 3200 ISO as well, but 1600 is a reasonable value in terms of sensitivity and retained dynamic range. Some swear by 800, I usually go for 1600.

--
My equipment: Nikon D3s, D800, EOS 450D
The ladies equipment: Panasonic LX5 and GH2 for video
 
Did you use a battery pack or a power adapter?

Batteries heat up as they discharge.

Also were these 2 darks done at the same temperature?

Did you do a series of darks at ISO800 before you did the ISO 1600? In which case you probably are only looking at increased thermal noise (dark current) from the previous ISO800 set.

ISO 800 or ISO1600 does not mean a lot really. All ISO is on these cameras is the amplifier gain. You are not gaining any extra signal here only amplification of the same signal.

So if you shoot in RAW. Take 10 darks and sigma reject combine to make a master dark and use adaptive darks in a program like Images Plus (the darks will be different to your light exposures because they will most likely be taken at different temperatures).

For adaptive darks to work you also need to take a bias image, ie a snapshot of the electronic and fixed pattern noise of the camera - typically fastest shutter speed, 6 or 8 and sigma reject combine them).

These darks look OK to me.

Keep in mind typical CCD performance is dark current (thermal noise) doubles every 6-7C increase in sensor temperature.

Greg.
 
See my comments inline:
Astrophotographer 10 said:
Did you use a battery pack or a power adapter?
Click to expand...
Battery. I know this should be taken into account. Power adapter not yet in da house.
Astrophotographer 10 said:
Also were these 2 darks done at the same temperature?
Click to expand...
They are part of a series. Yes, same temperature, in the house, 18 degrees.
Astrophotographer 10 said:
Did you do a series of darks at ISO800 before you did the ISO 1600? In which case you probably are only looking at increased thermal noise (dark current) from the previous ISO800 set.
Click to expand...
No, that's not the case. I also see the same behaviour using one darkframe only. I know the influence of heating the sensor, so I left some time between the 800 and 1600 series.
Astrophotographer 10 said:
ISO 800 or ISO1600 does not mean a lot really. All ISO is on these cameras is the amplifier gain. You are not gaining any extra signal here only amplification of the same signal.
Click to expand...
I know, but you do gain extra signal. Quite some years ago I tested all this and wrote an e-book about it :). You do get more signal from faint fuzzies at higher ISO. The trade-off is to find the optimal settings for a given camera.
Astrophotographer 10 said:
So if you shoot in RAW. Take 10 darks and sigma reject combine to make a master dark and use adaptive darks in a program like Images Plus (the darks will be different to your light exposures because they will most likely be taken at different temperatures).
Click to expand...
Of course. I am using ImagesPlus since version 1 :)
Astrophotographer 10 said:
For adaptive darks to work you also need to take a bias image, ie a snapshot of the electronic and fixed pattern noise of the camera - typically fastest shutter speed, 6 or 8 and sigma reject combine them).
Click to expand...
You are right again. And don't forget the flat field as well.

That is why the astro noise reduction on Nikon NX2 will only be good for casual shots. There's no way to inject all this into NX2. But is nice of Nikon for having thought about it for people making startrail images above landschapes. They usually don't care about darks, bias or flatfields.
Astrophotographer 10 said:
These darks look OK to me.
Click to expand...
To me too. I was just being curious about the jump in noise between 800 and 1600. And if I see this, one would fare a lot better keeping the ISO below the digital gain limit. Much less residual noise to expect.
Astrophotographer 10 said:
Keep in mind typical CCD performance is dark current (thermal noise) doubles every 6-7C increase in sensor temperature.
Click to expand...
Oh yes, that's why in summer I usually have a set of icepacks near the camera :)
Astrophotographer 10 said:
Greg.
Click to expand...
--
My equipment: Nikon D3s, D800, EOS 450D
The ladies equipment: Panasonic LX5 and GH2 for video
 
Samuel Dilworth said:
Hi Dominique. Why is your last image, quoted below, so much noisier on the left side compared to the right side? Did you put the 1st and 12th images together for comparison, or is this a sensor artefact of some kind?
Click to expand...
When inverted, would this not be the side of the sensor closest to the warm battery?
 
Dominique Dierick said:
That could be an explanation yes. Noise setttings were all identical...
Click to expand...
So it might not be necessary to drop all the way down to ISO 800 for astro-purposes - that ISO 1250 might give you better results?

I'm guessing that you're already planning on another test run for ISO 1250, am I correct? :) If so, I would be extremely interested in the results. Comparing them you would not only find out more about the sensor characteristics, but also come out with a simple rule of thumb for D800 astrophotography.

--
regards
Janne Mankila, Finland
 
If you look at the entire frame, there are two 'hot' zones, left and right, and a stroke at the bottom that you see pop up in regular photography at very high iso too (magentacolored). The dark zone is in the middle.

--
My equipment: Nikon D3s, D800, EOS 450D
The ladies equipment: Panasonic LX5 and GH2 for video
 
Conclusion, if you want to get the lowest long exposure, higher ISO noise for astrophotography, stop at 1250 ISO. As of 1600 ISO, likely the digital amplification adds to the noiselevels. And anyway, 1250 ISO is enough, given the characteristics of the sensor.

D.

--
My equipment: Nikon D3s, D800, EOS 450D
The ladies equipment: Panasonic LX5 and GH2 for video
 
Dominique Dierick said:
Conclusion, if you want to get the lowest long exposure, higher ISO noise for astrophotography, stop at 1250 ISO. As of 1600 ISO, likely the digital amplification adds to the noiselevels. And anyway, 1250 ISO is enough, given the characteristics of the sensor.
Click to expand...
An easy to remember and useful rule of thumb. Thank you. And I think a part of the thanks go, of course, to Bill Claff for pointing out the optimal ISO in his extensive studies on sensor behaviour.

Dominique, can we assume that, in general, ISO 1250 is enough for astrophotography? Seeing as it's only one third of a stop slower than ISO 1600, which is widely used. Added to the high resolution of the sensor, this could mean that D800 is finally an adequate astro-Nikon? :)

--
regards
Janne Mankila, Finland
 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top