ISO, invariance and dynamic range

[smr]

Hi all,

I am wondering if I should be using ISO 200 when imaging deep sky objects with my Canon 80D as there aren't really any definitive answers, and from what I have read mainly estimations and guess work. There is a website which shows which ISO to use for a lot of different DSLRs and their sensors, with the 80D being purportedly able to use an ISO of 200, which is tremendous really considering the low noise that such an ISO setting generates. But that website isn't working at the moment.

Then I saw some absolutely fantastic astro images on astrobin, taken with the 80D at ISO 1600. I messaged the author and asked him his thoughts but he hasn't logged in yet.

I have since been reading up and trying to understand what ISO setting really would be best to use and found some sensor stats on DXOMark. Here's some info on the sensor (80D on the left and 7Dmk2 for reference on the right)

What I would really like to know his how can it be estimated that the 80D works best at ISO 200 and in pushing the ISO up any further is detrimental to dynamic range? I don't understand how to ascertain that from these graphs.

Here's ISO graphs between the 80D and 7Dmk2 - the 7Dmk2 apparantly works best at an ISO of 1600 - how would I know this and that the 80D works best at ISO 200?

 

 

[Alien 13]

Another 80D user here, from what I understand the "best" ISO starts from when the graph is more or less a straight line so just below 1600 for the 7D 11 and 160 ish for the 80D. The theory behind it is that no extra noise will be added to the image by boosting the level in post processing compared to shooting at a higher ISO in the first place.

I think this is the other site you referred too. 

http://dslr-astrophotography.com/iso-values-canon-cameras/

I shall be watching this with interest, I personally shoot at ISO 200.

Alan

[smr]

5 minutes ago, Alien 13 said:

Another 80D user here, from what I understand the "best" ISO starts from when the graph is more or less a straight line so just below 1600 for the 7D 11 and 160 ish for the 80D. The theory behind it is that no extra noise will be added to the image by boosting the level in post processing compared to shooting at a higher ISO in the first place.

I think this is the other site you referred too. 

http://dslr-astrophotography.com/iso-values-canon-cameras/

I shall be watching this with interest, I personally shoot at ISO 200.

Alan

Yes that's the website, not working for me at the moment. What interests me as well is the ISO invariant at 509 value on the 80D graph.

[Alien 13]

I think that there are quite a few 80D owners on this site so hopefully they will chirp in.

A lots of Nikon cameras are ISO invariant and shoot at around 200 ISO for AP too so its perhaps not complete madness.

Alan

[rigradio]

I think you use the wrong graph.

This is the correct one: 

Unfortunately, it looks like the Canon 80D is "NOT as isoless as" Nikon 5300 which commonly recommended using ISO200 for AP

CMIIW

[smr]

According to this http://dslr-astrophotography.com/iso-values-canon-cameras/

ISO 200 is the correct value for the 80D. I just wondered how that was ascertained.

 

So what ISO should I be using then? 

[wxsatuser]

For faint objects it's useful to squeeze more tonal steps into the shadow area.
As ISO is increased the full well, dynamic range and read noise reduce.

For the 80D you can use ISO 400 instead of 200 and only loose about half a stop of DR but the read noise has dropped
from 4 electrons to nearly 3 electrons. Increasing the ISO gives better read noise and forces more tonal range into
the shadows. The shadow area is the hardest area to distinguish signal from noise and forcing more tones in there
will help when stretching.

What must be remembered as the ISO is increased the total exposure should not decrease.
The exposure at 400 ISO should equal the exposure at 200 ISO.
4 x 2minute subs at ISO 200 will equal 8 x 1minute subs at ISO 400 etc.

[Pompey Monkey]

5 hours ago, wxsatuser said:

For faint objects it's useful to squeeze more tonal steps into the shadow area.
As ISO is increased the full well, dynamic range and read noise reduce.

 

That's true, but also stacking a number of images increases the effective dynamic range greatly, providing the software supports the appropriate bit-depth.

[smr]

On 08/02/2019 at 04:41, wxsatuser said:

For faint objects it's useful to squeeze more tonal steps into the shadow area.
As ISO is increased the full well, dynamic range and read noise reduce.

For the 80D you can use ISO 400 instead of 200 and only loose about half a stop of DR but the read noise has dropped
from 4 electrons to nearly 3 electrons. Increasing the ISO gives better read noise and forces more tonal range into
the shadows. The shadow area is the hardest area to distinguish signal from noise and forcing more tones in there
will help when stretching.

What must be remembered as the ISO is increased the total exposure should not decrease.
The exposure at 400 ISO should equal the exposure at 200 ISO.
4 x 2minute subs at ISO 200 will equal 8 x 1minute subs at ISO 400 etc.

Thanks, so I should be using ISO 400 then?

Where do you read that ISO 400 equals 3 electrons? I need to read up on this stuff more and what electrons etc. means!

[wxsatuser]

You can use 400 as the read noise is lower and only loose half a stop of dynamic range.
Is it the best, I don't know, give it a try and see if there is a difference.

Go to here pick 80D and hover over the points to get the read noise.
http://www.photonstophotos.net/Charts/RN_e.htm

[smr]

The read noise decreases the higher the ISO goes, and stacking lots of subs increases dynamic range does it not ? Then shouldn't I use an ISO like 800 or 1600?

[mvas]

Are you imaging from a very Dark Site or a Light Polluted Site?
In your stacked image, what percentage of your "Overall Noise" actually comes from "Read Noise" ?
If Read Noise is not a significant percentage of your Overall Noise, then why increase ISO, which reduces your Dynamic Range?
Have you calculated how many more exposures are required, to restore your lost Dynamic Range, each time you double the ISO?
When you double your ISO, and you don't shorten the exposure, then you risk clipping the bright areas to pure white <= not good!
More than 50% of the stars in the typical astro-photo should be yellow, orange or red, not pure white.
Objects like galaxies with bright cores and the Orion Nebula have a huge Dynamic Range.

One guideline suggests that Read Noise should be 5% or less of the Overall Noise ...

[smr]

47 minutes ago, mvas said:

Are you imaging from a very Dark Site or a Light Polluted Site?
In your stacked image, what percentage of your "Overall Noise" actually comes from "Read Noise" ?
If Read Noise is not a significant percentage of your Overall Noise, then why increase ISO, which reduces your Dynamic Range?
Have you calculated how many more exposures are required, to restore your lost Dynamic Range, each time you double the ISO?
When you double your ISO, and you don't shorten the exposure, then you risk clipping the bright areas to pure white <= not good!
More than 50% of the stars in the typical astro-photo should be yellow, orange or red, not pure white.
Objects like galaxies with bright cores and the Orion Nebula have a huge Dynamic Range.

One guideline suggests that Read Noise should be 5% or less of the Overall Noise ...

Bortle 6 light polluted skies from a village in a semi rural location. I'm not sure how much of the noise is from read noise. I've just bought a light pollution filter to help with general sky fog and glow so I'm waiting for the Moon to not be so bright to see how that helps. 

[coatesg]

In light polluted skies, for broadband (ie not narrowband!) pictures, ISO400 would be fine. 200 may also suffice - the likelihood is that sky background will greatly overwhelm read noise, as long as the exposures aren't too short. 

The subs may look rather weak, but stacking and stretching will reveal the target while not saturating bright stars to the same extent as for a higher ISO with smaller dynamic range.

[mvas]

59 minutes ago, smr said:

Bortle 6 light polluted skies from a village in a semi rural location. I'm not sure how much of the noise is from read noise. I've just bought a light pollution filter to help with general sky fog and glow so I'm waiting for the Moon to not be so bright to see how that helps. 

I see no reason to increase the ISO, which decreases your Dynamic Range and only slightly decreases Read Noise.
Check the peak on your Histogram
You need to expose long enough to get a clean separation of the Histogram from the left wall
The peak can be between 20% to 33%
This will determine your exposure time, under light-polluted skies.
Finally, you will need many exposures to reduce the noise from the Sky Glow.
Here is link to M31, from Bortle 6, notice it took John Rista 7.5 HOURS at ISO 400 using Canon 5D Mk-III ...
M31-Bortle-6
 

[smr]

2 hours ago, mvas said:

I see no reason to increase the ISO, which decreases your Dynamic Range and only slightly decreases Read Noise.
Check the peak on your Histogram
You need to expose long enough to get a clean separation of the Histogram from the left wall
The peak can be between 20% to 33%
This will determine your exposure time, under light-polluted skies.
Finally, you will need many exposures to reduce the noise from the Sky Glow.
Here is link to M31, from Bortle 6, notice it took John Rista 7.5 HOURS at ISO 400 using Canon 5D Mk-III ...
M31-Bortle-6
 

But as said further up. ISO 400 is 3 read noise compared to 4 read noise at ISO 200 and 400 ISO is only half a stop less dynamic range...

[smr]

I'm still really quite frustratingly confused as to which ISO to use... having had a good 10 minutes spent looking at Canon 80D images on astrobin there are a variety of deep sky images taken from different skies, some quite a bit darker than others, yet everyone seems to use either ISO 800 or 1600. I saw one image which used ISO 400. None use ISO 200. 

So it makes me wonder if I am using the wrong ISO. 

This for example taken at ISO 800...

https://www.astrobin.com/315425/B/?page=5&amp;nc=user

 

[Alien 13]

5 minutes ago, smr said:

I'm still really quite frustratingly confused as to which ISO to use... having had a good 10 minutes spent looking at Canon 80D images on astrobin there are a variety of deep sky images taken from different skies, some quite a bit darker than others, yet everyone seems to use either ISO 800 or 1600. I saw one image which used ISO 400. None use ISO 200. 

So it makes me wonder if I am using the wrong ISO. 

This for example taken at ISO 800...

https://www.astrobin.com/315425/B/?page=5&amp;nc=user

 

The only real way to find out is by shooting a couple of test images with exactly the same exposure times, one at ISO 200 and the other ISO 800 and post them here. You could perhaps even do it with a single sub and stretch them to the same brightness level (might have to be a linear stretch though rather than a curve).

Alan

 

[davew]

17 minutes ago, smr said:

I'm still really quite frustratingly confused as to which ISO to use... having had a good 10 minutes spent looking at Canon 80D images on astrobin there are a variety of deep sky images taken from different skies, some quite a bit darker than others, yet everyone seems to use either ISO 800 or 1600. I saw one image which used ISO 400. None use ISO 200. 

So it makes me wonder if I am using the wrong ISO. 

This for example taken at ISO 800...

https://www.astrobin.com/315425/B/?page=5&amp;nc=user

 

I think you are looking at the image details in isolation. By that I mean that you don't know why the image you posted the link to was shot at ISO 800. It could be he doesn't know the best settings either and was just copying someone else on the internet who in turn copied someone else .......

I've helped out at a few imaging courses over the last year or two and I can tell you most people use the ISO value that let's them see a picture on their rear screen. Nothing more nothing less.

I also think you wouldn't be able to tell what ISO was used just by looking at the image ! If he'd used ISO 400 or 1600 you'd never know. 

I'd do what Alan ( Alien ) has suggested. If you're imaging in a very dark place you might be pleasantly surprised but if you image under light pollution you'll wonder what all the fuss was

Dave.

[smr]

6 minutes ago, davew said:

I think you are looking at the image details in isolation. By that I mean that you don't know why the image you posted the link to was shot at ISO 800. It could be he doesn't know the best settings either and was just copying someone else on the internet who in turn copied someone else .......

I've helped out at a few imaging courses over the last year or two and I can tell you most people use the ISO value that let's them see a picture on their rear screen. Nothing more nothing less.

I also think you wouldn't be able to tell what ISO was used just by looking at the image ! If he'd used ISO 400 or 1600 you'd never know. 

I'd do what Alan ( Alien ) has suggested. If you're imaging in a very dark place you might be pleasantly surprised but if you image under light pollution you'll wonder what all the fuss was

Dave.

Yes but then with some targets, take the jellyfish nebula for instance. I imagine you wouldn't see anything in a single sub even at ISO 3200. But with M42 you can see it at ISO 200.

[davew]

4 minutes ago, smr said:

Yes but then with some targets, take the jellyfish nebula for instance. I imagine you wouldn't see anything in a single sub even at ISO 3200. But with M42 you can see it at ISO 200.

Then do what Alan suggested. Seriously. I would err on the lower side but until you test you won't know and I, for one, can't tell you.

As a fore instance. Loads of ISO invariant cameras ( Mainly Nikon ) are shot at 200 or above. Do you know why ? It's because some Nikons don't play well at ISO 100 and you wouldn't know that without testing. Nothing to do with read noise, DR or anything else. I know you use a Canon so that was just an illustration of getting the wrong ideas.

Dave

[smr]

22 minutes ago, Alien 13 said:

The only real way to find out is by shooting a couple of test images with exactly the same exposure times, one at ISO 200 and the other ISO 800 and post them here. You could perhaps even do it with a single sub and stretch them to the same brightness level (might have to be a linear stretch though rather than a curve).

Alan

 

I did do this, to the best of my abilities...

I followed a guide online and it was to shoot in a dimly lit room - manual exposure mode, same shutter speed and aperture on each image just changing the ISO each time, and then bringing them back to the same brightness in post...

 

ISO 100                                                                                                                               ISO 200

ISO 400                                                                                                                                ISO 800

 

 

ISO 1600                                                                                                                             ISO 3200

[smr]

I don't know why, or possibly because I don't know what I'm looking at, but ISO 100 and 200 look a grainier/worse than ISO 400 and 800.

[Alien 13]

Its difficult to tell from the images above, perhaps using a printed out "test card" type image that can quantify differences or maybe even something like an eye chart.  

Alan