Hit the DSLR quality buffer with IC434

[Twisted Lip]

Hi all, so had a very (surprisingly - given the recent weather) clear night last night so took the following of a good 'ol favourite IC 434 - Horses Head.

Even though the image came out ok I think the amount of noise demonstrates that I'm hitting the limit of what is possible with a DSLR.

Anyway, I'd like to hear your thoughts as always and open to all criticism as ever

10x900s @ISO 400 (few more subs were binned due to aeroplanes, etc)

5 darks, 10 bias, 10 flats. Stacked in DSS and processed (as best I could get out of it) in PS CS5

Cheers all, clear skies

Will

[wavydavy]

Your very critical of your pics........I think your pic of the horsehead is ace...! Good job.

[narrowbandpaul]

5 darks aren't really enough. You need to reduce the random component of the darks (which you do by taking a bunch and "averaging") so you will end up injecting the random component in to your corrected image.

To really maximise your DSLR you will need a very good calibration. A very good set of darks and flats and bias frames to remove all the possible noise sources.

Kudos for the use of 15min subs, this is helping a lot.

Overall it's a very nice image, but your calibration frames could certainly be improved. You need more flats and bias frames.

I know getting darks with a DSLR is very tough but you eek out the last but of detail you need excellent calibration.

You need more flats too, again as with all data in astronomy you can never have too many. More bias frames would also be beneficial.

You said you were open to criticism, and I hope this didn't appear too critical, but you posted saying you were near the limit, and I agree you are close but the use of better calibration will give you a step up, though it won't be dramatic.

The rule is data, data and if in doubt, more data.

Hope that was helpful

Paul

[Stargazer33]

Very nice image.

Are you powering your DSLR from a battery or via a power lead? I have heard a lot about the camera batteries warming up during long exposures and causing noise on the frames closest to the battery. Perhaps running via a power lead will also help to give you that extra bit of improvement. That is is you aren't running a cooled DSLR.

I'm thinking of doing this for my 1100D - next DIY project!

[Digz]

That's a great shot. Like you when I tried this object I felt that I was hitting the limit of a DSLR but need to revisit it to be sure.

[Twisted Lip]

Thanks all, much ly appreciate the great comments

@wavydavy - thanks chap, yes sometimes I think I'm a little overly critical of my images!

@paul, don't feel you're being too critical at all! In fact I think you're comments are spot on. Definitely agree with you on the darks. My usual minimum is the same number of darks as lights but I was scuppered last night by (of all things!) battery power! As it started to cloud over I made the bias and flats but then the camera died half way through the darks! I would have like to have gather more certainly.

@Bryan - the above probably answers your question, running on battery :/ top tip though I hadn't even thought about it heating up. Definitely with considering a power hookup.

@Lee - much appreciated. This year I'd love to take the step up to CCD but of course. The funds need to allow!

Thanks all

Will

[StuartJPP]

Lucky you, I started setting up at 2am this morning after the wind died down and after completing sett up, roughly 25 minutes later, it had turned so misty that I couldn't even see Polaris...

Throw the discarded subs into the pot and select a sigma stacking method, and see what you get. Why throw away good subs due to small imperfections? Of course if they are out of focus or really bad then don' t use them, you have got nothing to lose by trying to use them.

I think there is a lot more mileage in DSLRs than most people think...one big step you can do is to modify your camera, it will bring out a lot more detail in the HH region, and lots of other regions as well. Here are two images, one with an unmodified DSLR and the other with, I know I have got a "fast" scope, but I also have a rickety mount that limits me to at most 90 second unguided subs. The unmodified camera image is a lot rougher than the modified version because I had to stretch it further to get more detail out of it (plus it was early days of processing for me).

Personally I would up the ISO to 800 and reduce the exposure time so that you get more subs to work with, i.e. less discarded subs due to issues. Depending on what the histogram looks like of course. There is a sweet spot of exposure time vs. noise which varies according to your kit.

As NarrowBandPaul says, more subs will bring out more detail.

Both these images were taken from fairly darkish sites (i.e. not my back garden), but still with a fair amount of LP. Are you reaching the limit due to LP at your imaging location?

Unmodified Canon 7D (50 x 90 second Lights, ISO1600 @ f/4):

Horsehead and Flame Nebula by StuartJPP, on Flickr

Modified Canon 1100D (50x 90 second lights, ISO 800 @ f/4):

Horsehead and Flame Nebula by StuartJPP, on Flickr

[narrowbandpaul]

Nice images Stuart,

Why do you advise to increase ISO and shorten exposure time? There is a big myth surrounding ISO and that is increasing the ISO somehow gets you more data just because the image looks bright. If you point as scope and open the [removed word] for identical times you will get the same number of photons. That is a fact. I wouldn't advise shortening subs to get more of them. I advise using 15 mins where possible and getting more of them, along with the extra calibration that I was referring to.

Yes you need to have sufficient quantities for a good sigma reject but not at the expense of using shorter subs. Shorter subs and more of them will hit your SNR due to multiple reads.

There is a secondary effect of ISO and that is on the dynamic range and read noise. The read noise will fall as you up the ISO but the DR will fall overall as you saturate the A/D converter before pixel full well. Yes a lower read noise is what you want but dynamic range is important too. The effects are type specific and small enough that the only way too tell would be through testing.

I did test my 300D for full well, read noise and dynamic range and confirmed the above results. I hear that an ISO 400 is about max DR (again this is camera specific). You may find that ISO 800 gives you a lower read noise without hitting DR too much. I suggest you measure this as the effect will be small.

I suggest you stick with 15min subs, get as many as you can. Use more darks, flats and bias frames. The more the merrier. This will boost your SNR. As to whether ISO 800 will help, that needs testing. Fortunately it's not too hard

Hope that was interesting....the ISO debate is usually a good one!

Paul

[PhotoGav]

I have prepared a library of darks taken with the camera (a modded 60D) in the fridge. Sensor temp was reading around 12-14 degrees. I'm currently exposing subs (funnily enough of the horse head nebula, so very interesting to see the pics above) and the sensor is running at around 11 degrees. I've used the library on a couple of projects so far and it seems to work. I shot 49 darks for each of a selection of sub lengths.

Bias - I have a master bias made from 199 frames, this is a one off master used for all projects.

Flats - I shoot 39 of these for each set of subs using an iPad as a light box.

All these calibration frames do seem to eek out a fairly decent image from the DSLR. It's no CCD when it comes to noise, but we have the advantage of wide field of view and being one-shot, it's quicker to acquire the subs.

Stuart's superb images above show that it is possible with a DSLR, so don't give up quite yet. Unless of course this all simply justification to hit the buy button for an Atik.....! In which case, don't delay, you know you want to!!!

[PhotoGav]

Just to add... I shoot everything at 800 ISO. That seems to be a good compromise between noise and image. I certainly don't want to go higher than that.

[daverave83]

The horse head nebula this is a object i am trying to improve on, i took about 12 darks and 10 bias, out of about 10 shots i had about 3 which had trailing.

i was hiting the limit of my scope at 1.10min exposures on an slt 102 skyscan.

this was adjusted through photoshop, so would flats and more darks help bring out the nebulosity?

[Digz]

Maybe I should qualify my earlier post by saying I'm using a Nikon DSLR - I like a challenge!!!!

This image was taken in 2011 with my D200 mounted on my NEQ6 and 250p DS.

[lensman57]

Hi all, so had a very (surprisingly - given the recent weather) clear night last night so took the following of a good 'ol favourite IC 434 - Horses Head.

Even though the image came out ok I think the amount of noise demonstrates that I'm hitting the limit of what is possible with a DSLR.

Anyway, I'd like to hear your thoughts as always and open to all criticism as ever

10x900s @ISO 400 (few more subs were binned due to aeroplanes, etc)

5 darks, 10 bias, 10 flats. Stacked in DSS and processed (as best I could get out of it) in PS CS5

Cheers all, clear skies

Will

Hi Will,

I think that this is a very good  capture, last night I was testing a new moodded 1100d on the same target and the same iso, 400  but with 300s subs using an Apo triplet @F4.8, I ended up dumping the data, the stack of 21 subs was so noisy that I really could not do much with it and I took 13 darks, 31 bias and 31 flats using an el panel, I just find the sky very strange at the moment, a couple of weeks ago I had much cleaner results with 30s subs on M42 but  I am sure the sky was a lot cleaner. 

Regards,

A.G

[Psychobilly]

We done on the image...

As well as the calibration frame improvements mentioned by Paul I would keep adding to the data...

I found that each year I visited a subject I doubled the amount of time spent on it until I wasn't happy unless I had a min of 8 hours of subs and ideally 12 or more.... Although the last few years the weather hasnt played ball with a lack of decent consequtive nights around new moon to acquire that much  data especially as my Orion's belt win doe is only about 2 1/2 hours and it coincides with the House central heating flue - the heating needs to be off for an hour before as well to let it all cool down so Orion is not a popular target with the rest of the household ...

A lot of what appears to be noise will then turn out to be the fainter extended nebulosity infront of the HHN and in the Ha curtain  behind it..

I just had a look at some of my pics of the region from 2009 and might revisit the data...

Peter...

[russellhq]

That's a nice start to a good image, and as others have said, just keep getting more data!

Also, what's the sky quality like at your capture site? A dark sky does wonders for image quality!

[ChrisLX200]

The camera temperature certainly does increase as soon as you start taking exposures, when I was trying to compile a bank of darks in my (at the time unheated) shed, 5x 10min dark frames, the temp started at 18deg for the first frame and was 26deg by the last. Outside, last night, the camera temp was stable at 15 deg after 10x 10min subs (it was pretty cold out, about 1degC, and dewey). With that variation in internal temperatures it's difficult to get a valid dark frame because 8deg will make a big difference to the noise. A cooled camera with temperature control has the big advantage here - not just because the sensor IS cooler but because the temperature is more consistent. Interesting comment about the battery being a big source of heat, I must do something about that

This was my take on Alnitak region, using a Canon 70D (unmodded), and 5x 10min subs through a TV Genesis SDF. I had no darks or flats - I was going to take them at the end but forgot and left it too late, the clouds rolled in spots of rain meant a rapid retreat to the shed. That's why I tried to generate some flats tonight, and failed.

This is by no means the limit and I've seen far better images (albeit through larger aperture instruments). However, the temperature control and better utilisation of narrow-band filters means I'm abandoning the unmodded DSLR in favour of a dedicated astro CCD camera, so I'll never reach the limit of what is possible with a DSLR.

ChrisH

[ChrisLX200]

Oh, I meant to add - I generally always use a 2" UHC filter with the DSLR mainly due to the local light pollution. A simple demonstration of the difference the filter makes is shown below. A 10min exposure would otherwise result in orange mush that is a bear to get rid of.

[Twisted Lip]

Just to add... I shoot everything at 800 ISO. That seems to be a good compromise between noise and image. I certainly don't want to go higher than that.

Top tip, I usually stick to 400 but I may change that to 800 and see how things get on. I definitely agree with you in that I like a lot of sub-time - I just like to get data. Just a shame that the DSLR heats itself up. Best start saving!

Will

[Twisted Lip]

We done on the image...

As well as the calibration frame improvements mentioned by Paul I would keep adding to the data...

I found that each year I visited a subject I doubled the amount of time spent on it until I wasn't happy unless I had a min of 8 hours of subs and ideally 12 or more.... Although the last few years the weather hasnt played ball with a lack of decent consequtive nights around new moon to acquire that much  data especially as my Orion's belt win doe is only about 2 1/2 hours and it coincides with the House central heating flue - the heating needs to be off for an hour before as well to let it all cool down so Orion is not a popular target with the rest of the household ...

A lot of what appears to be noise will then turn out to be the fainter extended nebulosity infront of the HHN and in the Ha curtain  behind it..

I just had a look at some of my pics of the region from 2009 and might revisit the data...

Peter...

Thanks Peter - its a good start as you say and I definitely take your advice and get more data. By the time I'm set up and ready to go, assuming there is not cloud cover I can usually get a good 3 hours out of Orion before it disappears behind the big trees!

That's a nice start to a good image, and as others have said, just keep getting more data!

Also, what's the sky quality like at your capture site? A dark sky does wonders for image quality!

Not bad but not 100%. To the West I have Reading from which I get a lot of glow, especially as they've started to move to LED street lighting. I use an astronimik CLS clip-in without which I wouldn't able to get 4 or 5 minutes without LP saturation.

Oh, I meant to add - I generally always use a 2" UHC filter with the DSLR mainly due to the local light pollution. A simple demonstration of the difference the filter makes is shown below. A 10min exposure would otherwise result in orange mush that is a bear to get rid of.

Saturday night was very cold here and I took some 10 min subs of M45 to measure the camera temp but even with the outside temp of close to 0, the camera was still warming up too much :/

[Stardust]

I have just had my first go at this one and still think it's amazing we can get what we have with a dslr.

I've just bought a modded 1100D and was advised to use 1600 iso, apparently it has better signal to noise ratio than at the lower iso speeds.

Personally I'd have gone with 800 but will give it a try.

[Stephen]

Good shot I've not read all the above replies but I would suggest

ISO 800

20 darks, flats and bias each

[Twisted Lip]

Thanks Stephen, weather allowing I'll definitely try and get more data on it and see how much further I can push the canon.

Will

[russellhq]

I have just had my first go at this one and still think it's amazing we can get what we have with a dslr.

I've just bought a modded 1100D and was advised to use 1600 iso, apparently it has better signal to noise ratio than at the lower iso speeds.

Personally I'd have gone with 800 but will give it a try.

Agree, that's the best ISO setting for the 1100D to minimise camera noise. But if you want to take long exposures >5min in urban areas you're going to quickly saturate the image. I used to image with 1600 now I just use 100 and take 15min exposures. But it needs to be cold outside to keep the temps down as the camera works best below +15C.

[narrowbandpaul]

I have just had my first go at this one and still think it's amazing we can get what we have with a dslr.

I've just bought a modded 1100D and was advised to use 1600 iso, apparently it has better signal to noise ratio than at the lower iso speeds.

Personally I'd have gone with 800 but will give it a try.

I would avoid relying on that kind of hearsay unless you know the source to be very reliable. These things can be easily measured and you will find what works best for your camera. True the read noise might be lower at higher ISO which is good but the full well is reduced too, which impacts dynamic range.

Cheers

Paul

[Xplode]

Cold weather is the DSLR imagers best helper for lower noise images.

On my 6D 3200 iso is fully usable, but it's much colder here in Norway than i the UK which makes a large difference.

Something else that helps a lot is using dithering to remove fixed pattern noise.

Here's an image i took back in november, click on it for larger version