What ISO / exposure time combination do you prefer?

[Juri]

New DSLR cameras are capable of high ISO setting with reasonable noise when it comes to "normal" photography. I've seen people here saying they use short exposures and high ISO settings, this seems a bit strange since the noise that is produced by high ISO settings is random, and cannot be removed in post processing without losing detail in image.

I understand that if you don't have a mount with tracking, you can't use long exposures but still, I think a stack of pictures with a combo of ISO 3200 and 0,8 s exposure might result in poorer image quality than ISO 1600 + 1,6 s exposure.

Am I wrong?

[Pete Presland]

its question of balance. i would have thought, find some nice dark skies and see what length exposure you can get without trailing. most people use ISO around 800

[Jannis]

, this seems a bit strange since the noise that is produced by high ISO settings is random, and cannot be removed in post processing without losing detail in image.

Am I wrong?

Yes, this is somewhat wrong. It's exactly the thing that it's completly random that makes it possible to remove it later by stacking many frames together. The noise is random and therfore gets averaged out, while the target subject is in all of the frames and stays when averaging all the frames together.

The thing about using higher ISO comes down to many things though, such as tracking, airplanes, satelites, birds, cars and so wich might cause the 500s exposure to be trashed. Then it's in many cases better to take 5x 100s frames at higher iso and stack 4 of them as one of them got trashed by a car passing by with full lights on directly at your scope. So, you're left with 400s total exposure fully usable, instead of one trashed 500s exposure.

It also comes to your mount. You might not even need a tracking EQ mount for the brightest objects if you just use the highest analog ISO the camera have and stack enough frames.

I've personally put this to the test very recently on both M45, and the horsehead nebula. Horsehead nebula is said that it's no point in even trying at with an unmodded DSLR if you can't get 1+ min exposures, but i was able to easily capture it with 5s exposures at ISO 3200 F/2.8. Sure, it's noisy and not the most detailed, but the 4x6" print i made of it looks wonderful. http://stargazerslou...3/#entry1797702

But you can't always "cheat" by just stacking short frames instead of taking long exposures though, as there will always be a minimum exposure requierd to capture all the faint details you want in the final image. Technically it's all about capturing as many of those photons as possible, and then stacking to lower teh background noise.

But in the end it all comes down to a balance of tracking time, chance of trashed frames, and level of noise at different ISO of the camera being used, as bunnygod1 said.

[dph1nm]

Remember that changing ISO does NOT alter the sensitivity of your DSLR to light. So you cannot reduce the required exposure time by going to higher ISO.

NigelM

[RikM]

I generally use 5min exposures at ISO1600 and f/5. I know the ISO doesn't effect the sensitivity, but I find the postprocessing easier if I use ISO1600 rather than ISO800. I don't have to stretch the image so far and can tone down the remaining noise more easily. It doesn't tend to blow the colours either on my unmodded 1000D. If I struggled to retain star colour, I would reduce the ISO to give more headroom on the dynamic range, but I don't find this necessary with my particular camera.

[rfdesigner]

High ISO = high gain between the sensor and the analogue to digital converter (ADC). The ADC has noise too, so a low gain means your image will look a little more noisy, but not much more until you get to very low gains (low ISO)

It is possible to measure this, and it's interesting that the noise levels don't get harmed much until you get to quite low levels, so you'll see people imaging as low as ISO400, even if they have ISO6400 available.

What you get by going for the lowest ISO you can before noise starts eating into your sub-frame is greater ability to image the bright areas without saturation, so you can take longer sub-frames which means fewer frames to stack and so fewer 'lots' of readout noise.

So..

100 x 1 minute at ISO3200 => 100x 5e readout noise => 50e

25 x 4 minute at ISO800 => 25x 7e readout noise => 35e

in this example both series of sub-frames will saturate at the same level, to make things completely even.. high ISO doesn't win!

hope this helps

Derek

[Juri]

Remember that changing ISO does NOT alter the sensitivity of your DSLR to light. So you cannot reduce the required exposure time by going to higher ISO.

NigelM

What the...? Newbie question: I guess this applies to astrophotography only?

[Robny]

Remember that changing ISO does NOT alter the sensitivity of your DSLR to light. So you cannot reduce the required exposure time by going to higher ISO.

NigelM

What the...? Newbie question: I guess this applies to astrophotography only?

So I'm new to Astro-Imaging but have been reading a lot, but that has also bamboozled me a little. Being in the photo trade myself I find it hard to get my head round and a little confused by the above statement.

[Robny]

Remember that changing ISO does NOT alter the sensitivity of your DSLR to light. So you cannot reduce the required exposure time by going to higher ISO.

NigelM

Sorry! This was supposed to be my quote

[themos]

Think about it! Why would you ever want to turn down the sensitivity of your sensor if it were a variable? You can control the amount of light with the shutter and the diaphragm so why would you ever use the sensor at less than its maximum sensitivity?

[Juri]

Think about it! Why would you ever want to turn down the sensitivity of your sensor if it were a variable? You can control the amount of light with the shutter and the diaphragm so why would you ever use the sensor at less than its maximum sensitivity?

Ummm, because of the noise? Again, noise gets worse as ISO get higher. Astrophotography may be a world of it's own when it comes to affects of noise when stacking images as Jannis pointed out. But still I'm suspicious...

[themos]

No, noise gets better as ISO gets higher! Ok, the improvement is not great but you will see, time and again, read noise (given as uncertainty in number of electrons) is lower at the highest ISO than at middle ISO.

The reason that high ISO has drawbacks is the reduced dynamic range: if you expose for 5 minutes, no matter what ISO setting you use, you will end up with the same number of electrons in each pixel site. That's ignoring the fact that photon arrival is random so you'll not get exactly the same number of photons but let's assume that we "replayed" exactly the same photon stream. That same number of electrons will be turned into a low or a high number in your image file, depending on the ISO setting. There's hardly any reason to use a high conversion ratio, so that we get say 5 units for each electron, because there's no way we'll ever get half an electron. All we are doing is using up 5 units when 1 would do and there's only so many in an image file. At high ISO, that is high conversion rate, we run the risk of not being able to record accurately a pixel that is as full of electrons as it can. That's losing dynamic range.

[rfdesigner]

Try reading this

http://www.astrosurf.com/~buil/50d/test.htm

You will notice the noise in ADC values (or ADUs) gets more at higher ISOs, but the noise in electrons (e) gets very slightly less. This is because the noise in electrons is magnified by larger values at higher ISOs. It's the noise in electrons that matter, so the sweet spot is lowest ISO we can get before the noise in electrons gets too much.

i.e. An ISO400 shot with noise of 10ADUs and an ISO800 shot with noise of 20ADUs both contain the same amount of noise... BUT the ISO800 shot will only have half the dynamic range of the ISO400 shot.

Derek

[Luke]

I normally use 5 minute shots at 1600 ISO.

I experimented with 800 ISO versus 1600 ISO and there didn't really seem to be anything in it.

Although the !600 ISO shots were brighter, they also had brighter noise, and it didn't really work out any different to me taking the 800 ISO shots and adjusting them to be as bright as the 1600 ISO shots.

The ISO as I understand it doesn't make your camera any more sensitive to light, able to pluck out fainter details, it just applies a "gain" that brightens up the image, including noise.

I try to avoid overexposing an image (detail lost due to clipping) or underexposing it (detail lost in nasty, streaky, low level noise).

[themos]

I experimented with 800 ISO versus 1600 ISO and there didn't really seem to be anything in it.

Except you lost star colour information for the many stars that were saturated in ISO 1600 but not in ISO 800.

[Luke]

Except you lost star colour information for the many stars that were saturated in ISO 1600 but not in ISO 800.

I disagree. Star colour is still there, if not at the centre, then at the edge, and it's easily added back. I prefer 1600 as I suspect I might lose some detail to read noise, but I don't think there would be that much in anyway based on my little bit of testing. I've seen plenty of great images at both ISOs. Experiment, don't lose too much sleep over it, and use what works best for you is my view.

[RikM]

I find the same. The star colour is retained even at ISO1600 with my camera. Only on the very brightest stars (Pleiades) would I need to drop the ISO to avoid burnout. Older cameras might have been less forgiving in this respect but I haven't found any difference in star colour at ISO400, ISO800 or ISO1600 other than on naked eye brightness stars.

[themos]

The DxOMark tests show it clearly for the EOS 500D

1. Dynamic range at ISO 400 11.11 Ev
   
2. Dynamic range at ISO 800 10.53 Ev
   
3. Dynamic range at ISO 1600 9.78 Ev
   

[RikM]

Certainly dynamic range is reduced at higher ISO, this is clear. But with my own images it doesn't make enough difference to be noticeable in the finished image and the processing is easier with an ISO1600 capture because you don't have to stretch so far.

[rfdesigner]

I find the same. The star colour is retained even at ISO1600 with my camera. Only on the very brightest stars (Pleiades) would I need to drop the ISO to avoid burnout. Older cameras might have been less forgiving in this respect but I haven't found any difference in star colour at ISO400, ISO800 or ISO1600 other than on naked eye brightness stars.

Just for the record.. what sub lengths and focal ratios are you imaging with?

Derek

[themos]

If you look at a mono CCD camera like the Atik 16ic http://www.stark-labs.com/craig/articles/assets/Atik16IC_API.pdf, it has a fixed gain (ISO equivalent). It is set so that at full-well capacity, which is measured at around 13,000 electrons, the ADU value will still fit in a 16-bit number (0 - 65535). The gain is about 4 ADU for one electron. There is no point doubling the gain on it because at 8 ADU per electron, a filled pixel (13000 electrons) would translate to 13000*8=104,000 ADU which can't be stored. It would end up as 65535, same as 8192 electrons. So we would lose the ability to tell the difference between 8192 electrons and 13000 electrons, a loss of dynamic range. For similar reasons, after a certain ISO setting in DSLRs, there's no point going higher, unless you REALLY don't care about the pixels that become almost full.

[Juri]

At this point I would like to thank me for asking and all of you for answering

Keep 'em coming, I find this very interesting and informative!

[RikM]

Just for the record.. what sub lengths and focal ratios are you imaging with?

Derek

7 min subs @ f/5 ISO1600 unmodded Canon 1000D

5 min subs @ f/5 ISO1600 unmodded Canon 1000D

[rfdesigner]

Rik.. , nice shots

So you're still using fairly long exposures, which is the key. rather than just winding up the ISO to try and compensate for short subs.

anyway, ISO1600 seems to work in practice, duely noted. would still be interesting to see ISO800 vs ISO1600 of the same object on the same kit.. but thats one for another day.

Derek

[dph1nm]

What the...? Newbie question: I guess this applies to astrophotography only?

No, it is the same for any photography. It is just that daytime photographers usually have so many photons floating around they don't really care!

NigelM