ISO setting for untracked astrophotography?

A disadvantage with higher ISO is you get less dynamic range.

A disadvantage with too low ISO is your image can be in the same level as the sensor noise (I read something like that I read on RN Clark's site).

So take a pic and look at the histogram. Adjust the ISO such that the main curve starts past zero.

A lot of dodgy stuff happens down in the shadow areas (i.e. severely underexposed regions) of most cameras and that is especially true of many Sony cameras, including my own Sony A7S. Your example shots prove this - lots of noise and nasty colour shifts.

The position of the back-of-camera histogram is more important than the ISO used. Adjust your shooting parameters (including ISO) so the histogram is one quarter to one third of the way from the left. That will avoid most problems.

Mark

sharkmelley said:

sharkmelley wrote:
(...)

The position of the back-of-camera histogram is more important than the ISO used. Adjust your shooting parameters (including ISO) so the histogram is one quarter to one third of the way from the left. That will avoid most problems.

Mark

Click to expand...

Mark knows his stuff, listen to him ;-)

What I would add is that the iso25600 image probably had some noise reduction wizardry applied, whereas the lower iso shots didn't (to the same degree).

Here's a plot illustrating read noise (the one remaining variable since you kept aperture and shutter speed constant) versus ISO:

http://photonstophotos.net/Charts/RN_e.htm#Sony ILCE-6000_14

Based on this, I would say that going higher than ISO 6400 is pointless.

Since the dynamic range decreases with increasing iso (e.g. http://photonstophotos.net/Charts/PDR.htm#Sony ILCE-6000 ) I would experiment and look for the lowest ISO possible (with a maximum of 6400) that, as Mark said, still puts the main "bump" on the histogram one quarter to one third from the left.

Good luck and be sure to show us the results!

regards,

--
Greg Van den Bleeken
www.pbase.com/gbleek
vimeo.com/vdbphotography
Take photographs *you* want to look at. Take photographs you want to *look* at. (Ed Leys)

The first link with 'input-referred read noise vs ISO' curve seems to suggest that the read noise decreases as ISO increases..? Or is it better to have high input-referred read noise?

betelgeuse91 said:

The first link with 'input-referred read noise vs ISO' curve seems to suggest that the read noise decreases as ISO increases..? Or is it better to have high input-referred read noise?

Click to expand...

It's a good question.

Yes, the read noise does decrease as ISO increases. That's typical for most sensors except for so-called "ISO-less" sensors where read noise is more or less constant across all ISOs. But read noise is very rarely the dominant source of noise. Part of the reason for getting the back-of-camera histogram in the right place is to ensure that the noise from the background sky glow (technically the shot noise) is the dominant source of image noise. So there's no need to worry about which ISO gives lowest read noise.

Mark

--
Takahashi Epsilon 180ED
H-alpha modified Sony A7S
http://www.markshelley.co.uk/Astronomy/

betelgeuse91 said:

Hi, I use Sony a6000 + Canon FD200mm + regular tripod for untracked astrophotography and I am wondering what would be the best ISO setting for me.

So I ran a test by taking pictures with varying ISO, having all other factors unchanged. Then I matched the total exposure of all pictures to the one of highest ISO(25600) in lightroom.



From left to right, ISO 25600, 12800, 6400, 3200, 1600, 800.

Clearly ISO 3200, 1600, 800 don't look usable. It seems that ISO 6400 performed best to me. Although the above is a cropped picture, I actually compared the full sized picture for all ISO settings and found that ISO 6400 was indeed the best in terms of the grain level in the noise.

Another thing I would like to point out is that in the full sized picture, ISO 25600 gave the most natural color, and the other ones including 6400 gave more red color in their overall noises. Although ISO 25600 gave more grain in the noise than 6400, the contrast and color was better with 25600 because of the more red noise in 6400. The following is the comparison between 25600(left) and 6400(right). I think 25600 has better color and contrast.



I am wondering if I should go with ISO 25600 or 6400. What do you guys think? Any thoughts or advises will be appreciated! Thank you.

Click to expand...

One step before that Trying to get sharp stars with a 200mm lens on an APSc camera = 300mm full frame equivalent untracked is mission impossible.

You are not going to be able to get a decent image. Even a short exposure will show elongated stars.

Try a much wider lens like a 14mm.

Greg.

Astrophotographer 10 said:

betelgeuse91 said:

Hi, I use Sony a6000 + Canon FD200mm + regular tripod for untracked astrophotography and I am wondering what would be the best ISO setting for me.

So I ran a test by taking pictures with varying ISO, having all other factors unchanged. Then I matched the total exposure of all pictures to the one of highest ISO(25600) in lightroom.



From left to right, ISO 25600, 12800, 6400, 3200, 1600, 800.

Clearly ISO 3200, 1600, 800 don't look usable. It seems that ISO 6400 performed best to me. Although the above is a cropped picture, I actually compared the full sized picture for all ISO settings and found that ISO 6400 was indeed the best in terms of the grain level in the noise.

Another thing I would like to point out is that in the full sized picture, ISO 25600 gave the most natural color, and the other ones including 6400 gave more red color in their overall noises. Although ISO 25600 gave more grain in the noise than 6400, the contrast and color was better with 25600 because of the more red noise in 6400. The following is the comparison between 25600(left) and 6400(right). I think 25600 has better color and contrast.



I am wondering if I should go with ISO 25600 or 6400. What do you guys think? Any thoughts or advises will be appreciated! Thank you.

Click to expand...

One step before that Trying to get sharp stars with a 200mm lens on an APSc camera = 300mm full frame equivalent untracked is mission impossible.

You are not going to be able to get a decent image. Even a short exposure will show elongated stars.

Try a much wider lens like a 14mm.

Greg.

Click to expand...

agreed.. if we simplistically use the 500 rule, the exposure limit will be 1.5secs.. making iso discussions largely moot.

for the OP: a "rule of thumb" to prevent elongated stars is to divide 500 by your effective focal length.. ie 500/(200*1.6) = ~1.5secs.

hence, reduce your focal length and lengthen your exposure.. in my case i most typically am at about 15 secs for 20-24mm focal lengths, and depending on aperture i will be at iso1600-6400

In some other posts that I saw some time ago, I saw someone saying that there are two types of noise to worry about: read noise and photon noise. By shot noise did you mean the photon noise? Or are they different things?

I really wish I had the link for that post but I don't unfortunately. But the person also mentioned that the photon noise can be flattened and removed by stacking many images but it is the read noise that is the problem we can't do anything about.

I will be stacking hundreds of images with about 1'' exposure. For this case, may I ask what your opinion is about the above person's comment? And which type of noise will be dominant?

Thank you for the comments.

betelgeuse91 said:

In some other posts that I saw some time ago, I saw someone saying that there are two types of noise to worry about: read noise and photon noise. By shot noise did you mean the photon noise? Or are they different things?

I really wish I had the link for that post but I don't unfortunately. But the person also mentioned that the photon noise can be flattened and removed by stacking many images but it is the read noise that is the problem we can't do anything about.

I will be stacking hundreds of images with about 1'' exposure. For this case, may I ask what your opinion is about the above person's comment? And which type of noise will be dominant?

Thank you for the comments.

Click to expand...

Now I have a better understanding of what you are trying to do - lots of very short exposures!

Shot noise and photon noise are the same thing. With such short exposures, read noise will definitely be a significant factor, possibly the dominant source of noise. You certainly want to use a very high ISO. Both photon (shot) noise and read noise are random and the so the signal-to-noise ratio will improve in the same way with stacking, whichever noise source is dominant.

I've seen some very impressive results from this technique. However, I'm sure you are already aware that a single tracked exposure of 100sec will have less noise than 100 stacked exposures of 1sec because you only have one lot of read noise.

Mark

--
Takahashi Epsilon 180ED
H-alpha modified Sony A7S
http://www.markshelley.co.uk/Astronomy/

I tried to use shorter lenses but they couldn't show the details of the objects well. I used 25mm last time for orion nebula, but it really didn't show anything. I unfortunately have to go with 200mm...

I should have been more clear about what I was trying to do. My apologies. I would love to have expose for long time too but the tracker is too expansive for me...

If both shot noise and read noise are random and can be flattened out, then all I should care about is the histogram position and the dynamic range I suppose?

betelgeuse91 said:

I tried to use shorter lenses but they couldn't show the details of the objects well. I used 25mm last time for orion nebula, but it really didn't show anything. I unfortunately have to go with 200mm...

Click to expand...

1-second exposures untracked with a 200 mm lens:

Stars move 15 arc-seconds per time second. On a camera with 4-micron pixels and 200 mm lens, the pixel spacing is 4-arc-seconds, so a 1 second exposure will be about a 4-pixel smear.

Maybe use a 100 mm lens, or with the 200 mm lens 1/2 second exposures.

Or build a barn door tracker for a few dollars.

betelgeuse91 said:

I should have been more clear about what I was trying to do. My apologies. I would love to have expose for long time too but the tracker is too expansive for me...

If both shot noise and read noise are random and can be flattened out, then all I should care about is the histogram position and the dynamic range I suppose?

Click to expand...

You won't get the histogram up to 1/4 or 1/3 from the left. I was assuming you were taking long exposures. I should have read the subject line of your original post: "untracked"

Mark

betelgeuse91 said:

I tried to use shorter lenses but they couldn't show the details of the objects well. I used 25mm last time for orion nebula, but it really didn't show anything. I unfortunately have to go with 200mm...

Click to expand...

No problem it just means you need a tracker. There are a few on the market and some are quite cheap. They are relatively easy to setup and align.

A longish exposure with a 200mm lens will require an accurate alignment with the celestial pole though as well as a reasonably accurate tracking mount.

Astrotrac 360 seems to be the most accurate.

I use a Vixen Polarie and its quite good with short focal lengths and around 200mm is about as long as I would use it for. Perhaps 300mm if I make sure the alignment is spot on.

I think you are setting yourself up for a loss. I can't see hundreds of 1 second images being practical. Firstly the processing and handling time will be large.

A simple tracker is not hard to use and will get good results. Stacking large numbers of 1 second images at high ISO (what camera are you using by the way?) sounds painful.

Greg.

That is actually what I am struggling with exactly. It takes a whole day to process those number of images... but I am not sure about investing on astrotrac-360. With this price, I could get a decent GEM mount with a motor drive. Thank you for the comment though.

betelgeuse91 said:

In some other posts that I saw some time ago, I saw someone saying that there are two types of noise to worry about: read noise and photon noise. By shot noise did you mean the photon noise? Or are they different things?

I really wish I had the link for that post but I don't unfortunately. But the person also mentioned that the photon noise can be flattened and removed by stacking many images but it is the read noise that is the problem we can't do anything about.

I will be stacking hundreds of images with about 1'' exposure. For this case, may I ask what your opinion is about the above person's comment? And which type of noise will be dominant?

Thank you for the comments.

Click to expand...

I wondered what ISO to use also. And what I did was to do some experiments and see for myself what I liked better. I also took enough to stack each ISO to see how I liked it. Now different cameras are going to produce different results, but my experience tended to show that the increase in noise tended to negate the extra sensitivity. Bottom line was that with the camera I did this with, I found I got better results using ISO 1600 rather than ISO 3200 and ISO 6400.

With short exposures, it is hard to produce great results, but stacking does enhance what you get. I am going to place a single image of using a full frame camera with a 300mm lens at f2.8 and ISO 1600, which is equivalent to doing this with a crop sendor and a 200mm lens, and then the result of stacking about 40 of those images and coming up with enough improvement that I could actually crop it a bit. This is not the type of shoot I normally do as I favor longer exposures done with tracking.

Here is one shot of the Hercules Cluster, a 2 sec shot.



And here is the results of stacking and stretching about 40 images using DSS for the stacking. It is cropped since the globular cluster is quite small. The result is an improvement, but nothing to write home about. And yes, stacking did reduce the noise and improve the resolution.



Here is one more just to give you additional ideas on ways to go about this. I like to experiment with different lenses and telescopes. This image was also the Hercules Cluster, m13, taken with a Atik Infinity (mono version) and using a Rokinon 135mm f2 lens and is a stack of six 46 sec images for a total exposure time of 280 sec. This was done using Atik's Infinity SW package that stacks on-the-fly as you observe. The Atik Infinity has a smaller sensor, thus makes objects look more magnified.

betelgeuse91 said:

That is actually what I am struggling with exactly. It takes a whole day to process those number of images... but I am not sure about investing on astrotrac-360. With this price, I could get a decent GEM mount with a motor drive. Thank you for the comment though.

Click to expand...

An Astrotrac or Fornax LighTrack can track better than any GEM up to the $4000 price range without resorting to autoguiding....


200mm is still quite short a fl , so you don't need to go to that extreme


A $289 iOprtron SkyTracker can handle 200mm as long as you don't push too much past a 1 minute exposure .


But why are you stuck on long focal lengths ?


These were all taken with just an 85mm lens.

For simply shooting the night sky while in a dark sky location, I find that 4"/ISO 500/ and F3.5 work quite well with my Nikon D5600. Most gratifyingly, the red and orange hues of some stars is plainly visible, although with blue stars the color differences become more vague since the atmosphere gives most stars a bluish cast anyway.

Unfortunately it's only on my phone that the pictures look this good, which is why I'm not sharing them right now.

ETA: I changed my mind, so I'll share one astrophoto anyway.

Dark Penguin said:

although with blue stars the color differences become more vague since the atmosphere gives most stars a bluish cast anyway.

Click to expand...

??? The atmosphere reddens star color. Atmospheric transmission decreases in the blue relative to red, and the effect becomes redder through more atmosphere. That is why the Sun is redder at sunrise and sunset.