DSLR versus CCD

[Darth Takahashi]

Today lots of people are using a DSLR to image the sky instead of a dedicated cooled CCD. I believe the reason is simple; if most of us came home to our wife's and told them that we had just dropped 4 to 10K on a camera that can only be used on clear moonless nights and can't take pictures of the family, well, i think you know where I'm going .

Canon currently offers many camera that are capable performers where imaging the heavens is concerned, however, once you have your camera and start taking images, determining the optimum settings can be difficult. I too here was apparently heading down the wrong path, striving for longer and longer single exposures.

Lots of people here in this community advocate lots of shorter exposures but with no real clear reason why, now I believe that I might of found the proof that shorter is better; I'm uploading a pdf by Craig Shark who goes to great pains of make a fair comparison . In the article it becomes clear why shorter is better where DSLR's are concerned.

My be some of you have seen this before, for me and possibly many others its new information.

Neil.

DSLRvsCCD_API.pdf

[brendant]

Not seen this particular article before Neil, very interesting read, thanks for sharing.

Brendan

[Sam]

That's a very interesting article, looks like ISO400 is about the right spot for a canon and loads and loads of shorter exposures.

[Darth Takahashi]

Yes, I agree, I'm now of the opinion that when light pollution allows I will only push for 6 min exposures max. Lots and lot of them of course, the question that's still open in my mind is whether or not to continue with DARKS, are they needed with exposures in the 3 to 6 min range at ISO400?

For safety I'll still do them but may be in he future I'll try stacking an image with and without them to see what the difference is?

Neil.

[brendant]

Interesting Neil as I have been debating the same issue as most of my recent images have not been dark subtracted and I really see little difference, maybe taking lots of lights with a DSLR and stacking gives just as good a result, DSS seems to do a good job of removing hot pixels in any case. Its not that I don't think they are a good idea but recently have not had the time to take and have not seen a difference. I generally shoot at ISO800 on the 400D (unmodded) @ up to 6 Min subs with heavy LP and can achieve acceptable results.

Brendan

[Psychobilly]

I don't bother with darks for the 1000D leaving DSS sort out the Hot pixels... but still use bias frames and now I have the light box fisihed for the Megrez 72 I'll be doing flats as well...

Mainly becasue of this... which is a room temp 20 min dark form a 1000D at ISO800... actually it was the 3rd of a set of 3x20 min subs taken with only 30s between them ...

Thats the biggest difference between the 1000D compared to the 350D - No visible "ampglow"

[rikyuu]

Sorry for being dumb :oops: (and I'm a bit under the weather at the moment), but I couldn't see the bit which says many shorter exposures are better than a long one, please can someone point it out to me .

I've thought about the question of long vs many short exposure for a long time (or many short times ), but never managed to conclude which is best.

Sorry if I'm hijacking the thread too, as the title is DSLR vs CCD, although some discussion is about long vs short exposure, so I'll ask my question about this here, hope that's ok;

I did a little experiment last night, taking a series of short exposures in a dimly lit room, stacking them in Registax and comparing them to a long exposure. It may not be a valid experiment as I'm not sure how it compares to astronomical photos, but no matter how many short exposure photos I stacked, the result was still a dim image compared to the long exposure image. I must be missing something fundamental, or perhaps I'm not using the correct stacking options. Or perhaps the idea is that many short exposure images reduces the noise when stacked, so the image can be brightened in Photoshop without increasing the noise. Please can someone clarify?

Many thanks

[moondog]

Or perhaps the idea is that many short exposure images reduces the noise when stacked, so the image can be brightened in Photoshop without increasing the noise.

rikyuu -yes - this is the reason!

8)

MD

[brendant]

Neil, sorry for the hijack and going off topic.

rikyuu :-

Its not just that many shorter exposures are better than longer ones there are many other factors which come into the equation. For most of us with light pollution, we will generally reach the sky fog limit long before exceeding our ability to guide or capture, the first thing you have to do is work out where this limit is, it might be 6 Mins @ ISO800 or say 10 mins @ ISO400, then to optomise your stacking so as to average out the noise you need a reasonable quantity of sub exposures, more is always better than less but again there will be an optimum number before you reach the state of diminishing returns whereby more subs does not yeild and improvement to the signal. It's all about the signal to noise ratio - a compromise, you need to expose for long enough to build up the signal but not so long as to swamp it out with noise and other issues such as light pollution, a fine balance that takes experimentation and is rarely the same for two people.

This is where CCD's score over DSLR's in that their sensitivity and signal to noise ratio is much better allowing them to capture light at a higher rate combined with low noise allows more detail and better resolution than your average DSLR, notwithstanding the excellent images that are possible from DSLR's but the exposure times are significantly longer for the same result.

Brendan

[rikyuu]

Thanks guys, always learning with this hobby.....

[George]

Also take into account your target, taking lots of short subs of a feint target....will reveal a feint target, obviously taking longer subs of that same target will capture more of the feint photons which will make your feint target a lot brighter. Astro photography is always about trade offs, you have to balance the capabilities of the camera, target brightness, LP & SNR.

My 350D has a sweet spot of around 5 mins@ ISO 800, my new 1000D which I'm just getting to grips with allows me the same 5 mins but @ ISO 1600 which allows me to capture more of the feint photons thus resulting in a brighter image with better SNR.

Also take into account ambient tempreture with a DSLR as this also effects SNR, a cold winters night will always be better than a warm summers evening.

As a comparison of ISO's here's the rosette I took tonight with the 1000D.

All 3 example images are RAW, no darks or flats and just a auto levels in CS3 and saved in jpeg format:

Rosette 5 mins @ ISO 400, moon light was interfering with this shot note the brightening to the left

Rosette 5mins @ ISO 800

Rosette 5 mins @ ISO 1600

[Psychobilly]

Still haven't had a real chance to use mine since its been modded and cleaned... another restless night lastnight with the gear setup in the obs "aligned" then clouded over... got up every hour to check but no decent breaks ...

Billy....

[dph1nm]

My 350D has a sweet spot of around 5 mins@ ISO 800, my new 1000D which I'm just getting to grips with allows me the same 5 mins but @ ISO 1600 which allows me to capture more of the feint photons thus resulting in a brighter image with better SNR.

Careful - the number of photons you get is determined by the exposure time, not the ISO. So 5min at ISO800 collects the same number of photons as 5min at ISO1600.

NigelM

[astro_dt]

Hi George

Thanks for posting the images that is a really useful illustration.

I use my 350 unguided at the moment but I am working on a tracking mount. Do you think all Canons have the same "sweet spot" if they were all using the same lens or will every one be different?

Cheers

Danny

[Darth Takahashi]

Sorry for taking so long to reply, I am actual home in the UK at the moment because my fathers sick and in hospital.

Anyway, back to the astronomy. I think one of the difficulties is in not believing that shorter exposure can capture the same detail as longer one's. The trick is to remember that the photon's are arriving randomly. Therefore, the shorter exposures can and do capture the same amount of detail as a single longer exposure. The only real difference in signal terms is the additional readout noise from each of the shorter subs.

Think about collecting rain in a bucket, you will collect the same amount of rain in the same time whether you leave the bucket for 30 mins or 6 x 5 mins. The difference is due to surface tension emptying 6 buckets mean each time you will lose some of your rain water and this can be equated to readout noise.

The other big benefit that I can see is with the darks. Darks should be as long as your lights and I hated capping the scope for 10 or 20 mins to capture a dart under the same conditions and time as the lights. Shorter darks means that I might doing them along with my lights as I go instead of when I'm ending an imaging session. However, depending on the camera and the noise you might also decide to skip them completely.

For me the article can be summed up as follows; Unless you can cool the imaging chip and gain better control of the build up of thermal noise, shorter is always better.

On balance after reading this article and listening to the opinions of others, I'm definitely going for shorter subs and not longer ones in the future.

And as George shows above, pictures worth 1000 words.

Neil.

[MartinB]

Also take into account your target, taking lots of short subs of a feint target....will reveal a feint target, obviously taking longer subs of that same target will capture more of the feint photons which will make your feint target a lot brighter.

This is certainly true for an unstretched average combined image. So, ignoring confounding factors such as shot and read noise, a single exposure of 30 mins may have a pixel value of 5000 whereas a 10x3 minute combine will only have a value of 500. However, that is of no consequence (other than read noise considerations). You could, if you wanted, just multiply each pixel value of the short exposure image by 10 to be at exactly the same value. Alternatively what we actually do is stretch which does a better job of making invisible differences in levels visible and brightening all at the same time.

To reinforce Neil's point, what actually matters is the accuracy of the relative pixel values. This is improved by increasing the length of overall exposure be it one long or several short.