General imaging questions

[Lognic04 (Logan)]

Hi all ,
Lately i have been told on other forums, for example in the case of my ngc 1097 that "with such a noisy dslr there is no sense in adding an extra 70 frames"
Honestly i think this is very confusing. I think what they were getting at was about diminishing returns.
But saying limit your total exposure time always to under 10 hours??
What if it's a very dim object and you want to pull out the faint bits of it? In my ngc 1097 i know it could use more exposure time, but i'm hesitant about trying to pull very faint dust out with a dslr with a lot of LP, i think i should do it again at a dark site.
What are your thoughts on this?

[Slawomir (Suavi)]

Hi Logan,

From my experience, measured noise n SNR quite closely follow theoretical patterns. To determine approximate total integration required to achieve desirable level of say background noise, one just needs to measure standard deviation in featureless background in an average sub, and then divide this number by a square root of the total number of planned subs. This will give a pretty good estimate of noise in a stack.

In my location and with my camera I get a st.dev of nearly 250ADU in featureless background in a few-minute Lum sub. So to achieve say 10ADU background st dev, I need 625 subs. If I was imaging from a darker site and had say st dev of 50ADU in the background in an average Lum sub, then I would achieve a similar result with merely 25 subs...that’s why I won’t try LRGB imaging anymore until we move to a darker location :-)

[Shiraz (Ray)]

every good frame helps Logan - the "diminishing returns" argument was fashionable for a while, but was wrong.

Adding excessively noisy frames (eg from a moonlit night) does not help, since the added noise overwhelms the benefit of the added signal - as a rough rule of thumb, a few subs with sky background greater than about 2x the level of the bulk of the subs will do more harm than good. If that is all you have though, use them.

doubling the sky background increases the noise by 1.4, but the signal remains the same, so the SNR is down by a factor of 1/1.4. The only way to get it back up is by extra integration - which adds more sky noise and dark noise..... If a dim galaxy takes 10 hours to a reasonable SNR under M21 sky, it will take ~25 hours to get the same result under an M19 sky. Sky brightness is a killer and the only way to defeat it is more integration.

[DJT (David)]

Interesting thread.
One related question I have is how do you know what an acceptable SNR is?
From Rays analysis I understand how to calculate a target ADU for a single sub based on a system and also what the most efficient exposure length for say an ASI1600 but without a target SNR in mind for a stacked image how would you plan for the number of exposures needed?

Or is it just a case of keeping on going until you are happy?

[Lognic04 (Logan)]

I have started to be under the impression that you just can't get a "good" image with a dslr (and many others are too) , because it's uncooled and has massive amounts of noise and relatively low QE.
Would you say this is true? Honestly while imaging and processing my data i have found this to be true. I think i would need to get thousands of frames just to get rid of this noise!
Or would imagine at a faster f ratio help to get more data? (when at a dark site)

[Slawomir (Suavi)]

I have never used a dslr for deep space photography, but have seen some spectacular images taken with DSLRs. As long as dslr is used at a dark site, it can produce some really good data. IMO Astrobin has plenty of great images taken with such cameras, I like this one for example: http://www.astrobin.com/full/307949/B/

[Camelopardalis (Dunk)]

I wouldn’t say that at all...folk put out some excellent images from DSLRs and OSC astro cameras.

Cooling helps (in summer), as does a dark sky to begin with, fast scope...there’s no single thing that leads to great images.

What I did (after moving to QLD) was to mod my 1100D, first to remove the IR-cut filter and later to add a cold finger. I would probably be less inclined to do it to a nice 700D like yours, but a used 1100D body can be picked up fairly cheaply if you are into experimentation.

[ZeroID (Brent)]

Canon 1100D, 1200D, 1300D all seem to use the same low noise sensor and a set of good darks has worked for me (1200D) to clean up any minor problems.
Having just gone to mono cooled I probably won't be using it much anymore but handy for wide field stuff.

[glend (Glen)]

I have an old Canon 450D, which is full spectrum, cold finger cooled, and it does well at dark sites. It is a bit power hungry because of the cooling, but once the sensor is below 0C the image is very smooth and noise free. You need to decide if modding a DSLR is for you. I believe the emergence of the relative lower cost CMOS mono cooled astro cameras like the super low noise ASI1600MM-C and the QHY variant, has put a dent in the DSLR mod world, simply because they are so good.

[sil (Steve)]

Originally Posted by DJT

Or is it just a case of keeping on going until you are happy?

This is enough for me. DSLR is not the best tool to use but for me its the best I have to use. So I try to understand the factors that effect my final post processed image and try to address those. Adding more frames is fine for me. I dont get hung up on number values of noise or star roundness or anything. I want to get colour as good as I can and pull out faint fuzzies as much as I can. I'm never going to win awards and I don't care about that. Getting the best (IMO) out of my data I can is satisfying. I'm constantly amazed at what humble consumer gear can capture of the universe hidden just beyond the edge of our eyesight.