Canon 6D - stacking short vs long subs?

[GTom]

I just upgraded to a 6D and checking in theory, what is the ideal ISO and frametime for the new setup.

Looking at the read-noise characteristics, ISO should not be kept low - however, increasing the ISO kills the full well capacity, at ISO3200 I got only 2400e-(!), which quickly leads to DR problems. I feel ISO1600 could be a meaningful compromise between FW and read noise.

Regarding noise and frametimes: I know, long subs are in general preferred. Looking at the overall S/N formula, in theory I'd loose only about 20% on the S/N if I shoot 10 subs instead of one long exposure  thanks to the ~3e- read noise.

Is my maths right? If the read noise is significantly below 5e-, can we really go in photon counting mode/save a lot on mounts?

 

 

I made a quick excel plot on the story: practically ISO3200 delivers the same results using 10 frames as 5 frames shot at iso 800.  I added 0.2e-/sec/pixel dark noise (~5°C cooled eos 6D value) and a signal, that generates 150electrons in a 2 minutes sub, I didn't add photon noise. 

 

[MARS1960]

Personally i try to use as low an ISO as the situation allows, i certainly wouldn't raise the ISO to 1600 if i can get what i need with twice the sub length at half the ISO.

I  always thought that the lower the ISO the higher the dynamic range, therefore the better the image.

Most noise can be removed in post processing and i use a crop sensor, your FF will have much less noise at a given ISO than my APS-C.

I have noticed that my images are now much better now i can take longer exposures and lower the ISO.

[GTom]

23 minutes ago, MARS1960 said:

Personally i try to use as low an ISO as the situation allows, i certainly wouldn't raise the ISO to 1600 if i can get what i need with twice the sub length at half the ISO.

I  always thought that the lower the ISO the higher the dynamic range, therefore the better the image.

Most noise can be removed in post processing and i use a crop sensor, your FF will have much less noise at a given ISO than my APS-C.

I have noticed that my images are now much better now i can take longer exposures and lower the ISO.

That's the nikon story, they got low read noise at low ISO.

BTW you can't exploit the full DR thanks the ADC, 6D is in reality capped at around 11EV, which it can still deliver at ISO1600.

[Knight of Clear Skies]

14 minutes ago, MARS1960 said:

I  always thought that the lower the ISO the higher the dynamic range, therefore the better the image.

You do get more dynamic range at lower ISO but there are a couple of problems - photons can be discarded (quantization error) and read noise is proportionally higher, so ISO setting is a bit of a compromise.

[MARS1960]

1 hour ago, GTom said:

I just upgraded to a 6D and checking in theory, what is the ideal ISO and frametime for the new setup.

Looking at the read-noise characteristics, ISO should not be kept low - however, increasing the ISO kills the full well capacity, at ISO3200 I got only 2400e-(!), which quickly leads to DR problems. I feel ISO1600 could be a meaningful compromise between FW and read noise.

Regarding noise and frametimes: I know, long subs are in general preferred. Looking at the overall S/N formula, in theory I'd loose only about 20% on the S/N if I shoot 10 subs instead of one long exposure  thanks to the ~3e- read noise.

Is my maths right? If the read noise is significantly below 5e-, can we really go in photon counting mode/save a lot on mounts?

 

 

I made a quick excel plot on the story: practically ISO3200 delivers the same results using 10 frames as 5 frames shot at iso 800.

 

Practically thats true but i would much rather (depending on object,LP,etc)take 5 subs of 12mins at ISO800 than 10 subs of 6mins at ISO3200.

[GTom]

25 minutes ago, MARS1960 said:

Practically thats true but i would much rather (depending on object,LP,etc)take 5 subs of 12mins at ISO800 than 10 subs of 6mins at ISO3200.

I completely understand, 2EV is a factor 4x in well capacity, which cannot be compensated by only 2x the number of subs (should be 4x). However, speaking of a DSLR, there is also the dark noise.

There is no perfect cooling for DSLR-s, even a cold-finger cooled unit could accumulate more dark noise during the high-temperature part of the shooting.

Further consideration is the mount. Of course, someone equipped with a super stable heavy mount, wind-protected dome, OAG, etc -> 20+min subs are not an issue. Put a 500mm lens on an astrotrac, ungided: you can

forget anything beyond 2-3minutes.

[alexbb]

And you may also take a look on this: http://dslr-astrophotography.com/long-exposures-multiple-shorter-exposures/

[GTom]

2 hours ago, moise212 said:

And you may also take a look on this: http://dslr-astrophotography.com/long-exposures-multiple-shorter-exposures/

Thanks, that looks much more detailed then my approach!

[StuartJPP]

On 13/12/2016 at 09:58, moise212 said:

And you may also take a look on this: http://dslr-astrophotography.com/long-exposures-multiple-shorter-exposures/

That is a good report and kind of confirms what I expected. Longer subs produce more noise but also captures more signal and that total exposure time is what counts ultimately (that was my take away). 

However the difference isn't massive, I think people spend too much time analysing rather than just getting on with it (good for those with time to analyse it for us). To me it would be better to just stick to something that works well for you which allows you to add data to previous sessions easily.

I limit my exposures to 5 minutes at f/4 at ISO1600 on the 6D. If this isn't optimal I am not too bothered as processing the image "properly" will have a much larger impact.

What isn't shown in those findings is the theory that taking 7200 x 1 second sub exposures will produce a very similar image. I believe that it won't...It will still be impressive what you can get from 1 second exposures but nowhere near as good as increasing the sub length to a specific point.

[Adam J]

This may help you find a balance if you are not aware of the site already:

http://www.sensorgen.info/CanonEOS-6D.html

Personally I would say that although you will get better S/N at higher ISO 3200 looks optimal from the charts, realistically you are limited in your choices by dynamic range so I would be tempted to keep it to 1600 or even 800 and go for the longer exposures. But its going to be target dependent. Something like M45 calls for more dynamic range than say the western veil as an example.

[Adam J]

On 12/12/2016 at 17:09, GTom said:

I completely understand, 2EV is a factor 4x in well capacity, which cannot be compensated by only 2x the number of subs (should be 4x). However, speaking of a DSLR, there is also the dark noise.

There is no perfect cooling for DSLR-s, even a cold-finger cooled unit could accumulate more dark noise during the high-temperature part of the shooting.

Further consideration is the mount. Of course, someone equipped with a super stable heavy mount, wind-protected dome, OAG, etc -> 20+min subs are not an issue. Put a 500mm lens on an astrotrac, ungided: you can

forget anything beyond 2-3minutes.

Here is some 20min cooled DSLR H-a for you mate not the best H-a of this target ever in the world but its possible to get the dark noisedown far enough and get those sub length without a dome on a humble HEQ5, mind you the 1000D used is not close to the quality of the 6D either.... 

Note this was with my RGB 1000D not the Mono modded one.

Highly Cropped.

[GTom]

Thanks Adam, that's a great shot! Proper cooling is certainly a witchcraft i have to learn...