Asking too much of a Sensor??

[Robny]

More curiosity than anything else....

As the title suggests, can you ask too much from a sensor.  More specifically a Canon 450d sensor (soon to be changed).  Regardless of how many lights I take,along with calibration frames I seem  to suffer from A LOT of noise once processed (using pixinsight).

So made me wonder if I am expecting too much asking my Canin 450d sensor to do 3 min exposure and if I would be better of taking shorter exposure times but more of them to get to the same integration time.

Reason I'm asking is because I'm somewhat dissapointed in the noise levels in my images despite stacking and calibrating.  I got 2hrs with of lights on M45 and I expected it to be better than it was after stacking and calibrating, I know 2hrs is not a huge amount but I just expected better

Cheers

Rob

 

[discardedastro]

DSLRs for astrophotography are always going to be more of a challenge than a proper cooled CCD/CMOS. An extreme case but I was imaging with an old Nikon D1x for a while on a 200PDS (so plenty of aperture) and the pattern noise even at low ISO was immense, making images very very hard to work with - it was basically impossible to calibrate out noise and leave a satisfactory frame, and integration even with plenty of dither couldn't achieve good results. If SNR is too low, you're always going to end up having to retrieve data from within or very close to the noise floor.

In the end I ditched it and bought an ASI183MM-PRO (similar to what you've done, if your sig's anything to go by). Haven't regretted it - it's night and day. Expect it won't be as dramatic an upgrade from a 450D, but still will improve.

[vlaiv]

Yes you are

In terms of sensor and noise reduction - most significant step that you can take is move to set temperature cooled one. This will let you calibrate your frames properly. Depending on your local conditions and sensor, dark current might be something that should not worry you too much (if other sources swamp it), but ability to properly calibrate frames is enormous advantage.

You also need to understand that SNR gains in terms of imaging time come as square law - each time you want to double your SNR you need to quadruple your imaging time. This means that additional hour of data will have different impact - depending on how much data you already collected. If you have one hour and add another one - total SNR increase will be by factor of ~1.4. If you have 16h of data and you add 1h more - total SNR increase in this case will be ~1.03 or just 3%.

This also means that if you have 1h of data - to double your SNR you need to shoot another 3h (that does not seem that much). If you have 8h of data, to double your SNR on that you need additional 24h hour of data - and that is a lot of imaging time.

Don't go for higher number of shorter subs if you can already shoot for 3 minutes - it will increase your noise (exact increase depends on host of parameters, but it will always be an increase if you use shorter subs).

If you shoot in light pollution - most significant step to get better SNR is to move to darker skies when imaging.

Next significant step would be to get set temperature cooled sensor.

[Robny]

Thanks for the responses guys, very interesting reading and I think you hit uppon a few points that effect me.

I do image in light polluted skies and getting to a dark sky site isn't that easy.  However when I get my new camera I'm also getting an IDAS d2 filter so that should help a little along with the cooled sensor.

So according square law, to double my SNR on a 2hr image would require a further 8hrs?

Thanks again

Rob

[vlaiv]

20 minutes ago, Robny said:

So according square law, to double my SNR on a 2hr image would require a further 8hrs?

Total of 8hrs, so 6hrs more.

Increase in SNR is the same as increase in SNR from having more subs to stack - when you add imaging time you are in effect adding more subs. SNR goes up like square root of number of subs - and it will in this case raise by ratio of longer total imaging time / shorter total imaging time - or 8 / 2 = 4, sqrt(4) = 2. So adding 6hrs does this.

[michael8554]

13 hours ago, Robny said:

can you ask too much from a sensor.  More specifically a Canon 450d sensor

I used to take my 450D exposures at 10 minutes quite regularly, with dither.

I never considered the results world class, but certainly not that noisy.

Michael

[Thalestris24]

Hi Rob

You might find this an inspiring web site. Dithering is a great help for reducing colour noise. Light pollution is always problematic - the sky background reduces your signal/noise significantly meaning you need to take a lot more subs to compensate. Don't forget to take sufficient calibration frames - about 30 of each is good. You can usually reuse most of them. How you do your stacking and post processing can make a big difference. Do post some of your images - people here are really helpful so I'm sure you will get some constructive advice  

Louise

 

[sharkmelley]

The crucial thing is to work out what is the dominant source of noise in your image.  Is it read noise, thermal noise or shot noise from the light pollution?   If light pollution is your main issue then you don't have a sensor problem. 

When calibrating using darks, make sure you have PixInsight's dark optimisation switched on - it does an excellent job of scaling the darks to match the lights, despite any temperature differences between the two.

Mark

[newbie alert]

What iso are you using, maybe drop the iso down 

[sharkmelley]

50 minutes ago, newbie alert said:

What iso are you using, maybe drop the iso down 

Unfortunately, the read noise is higher with lower ISO and banding artifacts will also be more intrusive.

Mark

[Robny]

I was shooting at ISO 800 originally, then I thought I would get better results by trying to drop it to 400, however...I guess based by the above comment that was never going to work, which it didn't.  Maybe I should try an higher ISO 1600 then based on that.

Hopefully after PAS all these problems will go away...and get replaced with lots more

[Demonperformer]

I have fairly heavy light pollution (was worse under the old orange lights) and the difference when I changed from a canon dslr to a cooled dslr was incredible. Apart from anything else, you can take proper darks. On an uncooled dslr, you can never be sure your darks are properly matched (in which case they may be adding more noise than they are removing). You are about to embark on an exhilerating phase of the journey!

[tooth_dr]

Hi Rob.

I would have used similar cameras (1000d and 40d) for some time.  For me 2 hours was never enough and I found the sweet spot to be 8 hours. I guess that’s no coincidence since it’s in ageeement with @vlaiv and his calculations.

I also never found lower iso to be any use, and I used 800 if I was shooting colour and 1600 if I had my Ha filter fitted.  I also found that lots of shorter exposures weren’t as easy to work with than fewer longer exposures. Typically I shot 300s or 600s or 900s subs.

Could you try that next time? 

 

[Robny]

9 hours ago, tooth_dr said:

Hi Rob.

I would have used similar cameras (1000d and 40d) for some time.  For me 2 hours was never enough and I found the sweet spot to be 8 hours. I guess that’s no coincidence since it’s in ageeement with @vlaiv and his calculations.

I also never found lower iso to be any use, and I used 800 if I was shooting colour and 1600 if I had my Ha filter fitted.  I also found that lots of shorter exposures weren’t as easy to work with than fewer longer exposures. Typically I shot 300s or 600s or 900s subs.

Could you try that next time? 

 

Thanks....good to know, I will try next opportunity but I'm soon going cooled osc.