been done to death but

[brrttpaul]

still cant get my head around exposure settings/gain with my asi 1600,  for some reason it just dont compute in my head whats the best for a deep sky image in bortle 5 skies, I have done youtube, tutorials etc but still end up scratching my head, sometimes my images are ok sometimes they are terrible. I tried the sensor analysis today but reading it means nothing to me, any pointers? heres a snapshot of the analysis

[rickwayne]

Have you watched Dr. Glover's talk? Made it pretty clear to me. 

 

[rickwayne]

Heck I think I'm gonna watch it again right now 🙂

[david_taurus83]

2 hours ago, brrttpaul said:

still cant get my head around exposure settings/gain with my asi 1600,  for some reason it just dont compute in my head whats the best for a deep sky image in bortle 5 skies, I have done youtube, tutorials etc but still end up scratching my head, sometimes my images are ok sometimes they are terrible. I tried the sensor analysis today but reading it means nothing to me, any pointers? heres a snapshot of the analysis

Now you've done the sensor analysis, that's the first stage. Next, you need to run Smart Histogram when your next out. Point at a dark patch of sky and measure the skies background with the SH feature. Just like sensor analysis, it will take a series of exposures at different gain values and work out the optimum exposure and gain value for your skies. It will offer a Unity Gain setting for shorter exposures and a Highest Dynamic Range setting for longer exposures but lower gain. Take it as a guide only but it's not a bad place to start.

[brrttpaul]

thx guys , I did plan to run the smart histogram next clear sky (so that should be june 2022) I havnt watched that video I must admit 

[windjammer]

Hi

I watched the video and I get the maths but don't understand it. The good doctor comes up with an exposure time with no reference to how dim or bright the object is.  Only the cam read noise and light pollution flux.  I don't agree LP is only shot noise,  more properly considered as a mean flux <p> plus sd Op.

What is wrong with exposing so the brightest bits just saturate and the LP lurks at the bottom ?  Then just take loads of frames and stack - though would be nice to compute when the marginal benefit of the next frame is pointless.

To go after the faint detail, overexpose the bright bits and hope you capture counts from the fainter bits before the LP comes chasing up from the bottom and saturation comes down from the top.

 

[spillage]

can I ask what software are you using to run the sensor analysis and smart histogram?

[JamesF]

21 minutes ago, spillage said:

can I ask what software are you using to run the sensor analysis and smart histogram?

I think that's part of SharpCap.  Possibly just the paid-for version, but I'm not sure.

James

[JamesF]

16 hours ago, windjammer said:

I watched the video and I get the maths but don't understand it. The good doctor comes up with an exposure time with no reference to how dim or bright the object is.  Only the cam read noise and light pollution flux.  I don't agree LP is only shot noise,  more properly considered as a mean flux <p> plus sd Op.

What is wrong with exposing so the brightest bits just saturate and the LP lurks at the bottom ?  Then just take loads of frames and stack - though would be nice to compute when the marginal benefit of the next frame is pointless.

To go after the faint detail, overexpose the bright bits and hope you capture counts from the fainter bits before the LP comes chasing up from the bottom and saturation comes down from the top.

My understanding of his view is that there comes a point, regardless of the brightness of the target, where the SNR cannot practically be improved upon and at that point you don't gain anything by exposing for longer -- effectively you're just "moving the histogram to the right" beyond that point.

James

[windjammer]

HI James

That may be his view, but that is not what his video is about.  In the video he asks what is the shortest sub exposure you should take to get  most of the noise benefit of a single exposure.  The single exposure SNR is best (he says) because there is only one hit of the cam read noise, whereas subexps have multiple hits.

By saying, OK, I will accept a subexp that is only X% worse in absolute noise level (not even SNR!) he then comes up with a value for subexposure time - based on cam read noise and LP noise.

It is elegant, granted, but according to this calculation you would use the same exposure length for snapping the sun as for your dimmest DSO!  

I do not have Sharpcap, and I am sure there is loads of good stuff in it.  But as to what exposure time to use on any given object I am stuck on the recipe I posted earlier.

On the topic of CMOS gain which he moves onto, I have no experience (CCD). But it does seem complicated!

Simon

[JamesF]

36 minutes ago, windjammer said:

It is elegant, granted, but according to this calculation you would use the same exposure length for snapping the sun as for your dimmest DSO!

I'm assuming that he's taking it as read that the exposure must stop before the sensor is saturated, but otherwise perhaps you would do just that given perfectly steady seeing.  The Sun, Moon and (some of) the planets are quite capable of saturating most sensors fairly quickly however, and the image scale might be five times or more what it would be for DSO imaging making atmospheric distortion more obvious, so we have to take a different approach.

I would suggest btw, that "the shortest sub exposure you should take to get  most of the noise benefit of a single exposure" is in effect the same as what I described as the point "where the SNR cannot practically be improved upon".  It may just be a matter of which way you look at the problem

It may be that he's wrong of course (I'm probably not qualified to judge), but for me the nice thing about him presenting his thoughts in this way rather than just hiding the process in some black box in SharpCap is that if he is wrong then people can point out where and why there are errors in his assumptions, the process can move forward and we all benefit.  For example, if you believe the brightness of the target matters then please do demonstrate exactly why that must be the case.  Once it's taken into account everyone should have the opportunity to create better images, which has to be a positive outcome.

For clarity, I'm not "rushing to Robin's defence" here.  I'm absolutely sure he can manage that more than adequately without my hindrance.  I do however value the opportunity to learn by debating the validity of the assumptions we make about how the imaging process works.

James

Edited November 26, 2019 by JamesF

[brrttpaul]

This is my problem where I went wrong  coming from a DSLR and knowing lower ISO is less noise I sort of looked at gain like ISO ( so I was thinking  gain @50 is going to be less noisy than gain@139)  and thats what keeps creeping in. (not only that but if you take a sample shot gain 50 and another in gain 300 say then gain 50 does look a lot smoother less noisy). I get now what the video says about once reaching a point theres no need to push it further I just dont get the bit where he says "so all you have to do is work out the maths this times that divided by that equals this.  I am in . I am in an orange zone using an equinox ed 80 and the 1600 mono cooled, so I think the ed 80 is roughly F6 thereabouts, Just roughly very roughly as a starting point I would imagine i should look at gain of 200 and expose for about 10 secs then? ( thats lum) then triple it for RGB)?

[david_taurus83]

What do your images look like at 139 gain? You can drive yourself crackers thinking about it too much.

[brrttpaul]

my images are a mixed bag really, you are right though it can drive you round the bend at times

[spillage]

Just from memory I think higher gain means you expose for longer but you need to be aware how dark you skies are. If you have poor skies then a lower gain with low exposure time should be used.

With narrow band you should use a higher gain. But do not take my  word on any of this but over at cloudy nights there are some monster threads about it.

Or you just set it to unity and adjust exposure times and ignore all the maths and equations.

For some reason I think I am using 200 gain and exposure time with L is about 120s but I still think this is too long.

[brrttpaul]

thx mark, im hoping to get some clear skies soon and do a few at different gains etc and see how it goes. I went from unity gain down to 50 and think i went the wrong way if im honest, 200 sounds in the ball park but shorter exposures i think

[Merlin66]

I use the ASI 1600 for solar imaging.....

I just stick with the Unity gain (139) (In FireCapture) and gamma 50.

The exposure set to give a good histogram +/- 80 %

Seems to work for me.

Ken

 

[spillage]

I sure its just a balancing game so try 50  unity and 200 gain but you will have to try different exposures on each maybe 30s, 60s and 120 (using broadband filters) . Don't bother doing any calibration frames and just throw the subs together and see how they look.

[brrttpaul]

Well there was a small chance last night so tried for the first time in ages to set up, I had gain at 200, exp for RGB was 15 secs. I tried to get an hours worth but when I ran it in blink i had to get rid of a load of bad subs. I am still having issues with SGPRO (keeps freezing if I take a single shot for framing) and also in sharpcap (I can do the sensor analysis but when I click on the brain in the histo to calculate correct exposure nothing seems to happen). right so set up didnt use a bat mask as wanted expecting to get anywhere slewed to a target (cant even remember which cluster it was and ran a sequence. stacked in pix and slight adjustment in LR, no darks or flats or bias. im happy with the result TBH so guess next stage now is to do a dark/flat library. reduced the temp also from -25 to -10