Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

ASI2600: Testing numbers not making sense

  • Please log in to reply
63 replies to this topic

#51 mewmartigan

mewmartigan

    Apollo

  • *****
  • Posts: 1,316
  • Joined: 02 Jul 2008
  • Loc: outside of Pittsburgh, PA

Posted 29 April 2020 - 08:56 AM

Marcus- Where did you order from? I'm guessing ZWO directly?


Originally Highpoint, but the one that arrived I ordered from ZWO direct. That seems to be the best option until they are readily available.

#52 mewmartigan

mewmartigan

    Apollo

  • *****
  • Posts: 1,316
  • Joined: 02 Jul 2008
  • Loc: outside of Pittsburgh, PA

Posted 29 April 2020 - 08:57 AM

I think three minutes at gain 0 with an f/7 scope might not be long enough. With the bayer matrix it has seemed to me like it is not as "sensitive" as I'm used to with mono cameras. In my experiments so far with an f6 scope at gain 0, 5min seemed decent but I thought I could push it to 10min. There might be a hint of star saturation at 10min though.

For me I want all the dynamic range and full well I can get, which means gain 0 for broadband.


Okay, I think I will get a set of 3 and 5 minute darks then and experiment if the clouds ever clear. Thanks!

#53 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 24,778
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 29 April 2020 - 11:53 AM

 

Test 1 - ASCOM Driver Gain=0
Read Noise = 3.469 e-
Full Well = 50,109
Dynamic Range = 14,443

 

Test 2 - Ascom Driver Gain = 23

Read Noise = 3.375 e-
Full Well = 38,235
Dynamic Range = 11,329

 

Test 3 - ASCOM Driver Gain=24
Read Noise = 3.347 e-
Full Well = 37,514
Dynamic Range = 11,209

 

Test 4 - ASCOM Driver Gain=25
Read Noise = 3.36 e-
Full Well = 37,301
Dynamic Range = 11,101

 

Test 5 - ASCOM Driver Gain=30
Read Noise = 3.344 e-
Full Well = 35,349
Dynamic Range = 10,572

 

Test 6 - ASCOM Driver Gain=100
Read Noise = 1.483 e-
Full Well = 16,274
Dynamic Range = 10,973

 

 

So, the dynamic range you have reported here is STEPS, not STOPS. Steps are the number of discrete steps of discernible data, given the noise. STEPS is a LINEAR measure of dynamic range.

 

STOPS on the other hand is a LOGARITHMIC measure of dynamic range. To convert steps to stops, you need to use a bit of math. You can do it pretty simply with base 2 log, but it can also be done fairly easily on a normal base 10 calculator. To determine stops, divide the full well in e- by the read noise in e- (NOT adu), take the base ten log (log or log10 on your average calculator), multiply by 20, and divide by 6. So, for Gain 0, you would have:

 

DRstp0 = log(50109/3.469) * 20 / 6 = 13.87 stops

 

For Gain 100, you would have:

 

DRstp100 = log(16274/1.483) * 20 / 6 = 13.46 stops

 

As you can see, the dynamic range of both gain settings is pretty darn close, both are well over 13 stops, which is EXCELLENT.

 

NOW, key thing to note here is that despite these two gains having the same dynamic range, they have different FWC levels. This is a very important factor, and it determines ALLOWED exposure lengths. If you are ok with shorter exposures, you can get the same results, with the same total integration time, by stacking more subs, at Gain 100. However, if you want the longest exposures you can get, then you might want to consider the lower gain, despite the higher read noise...as again, for the same total integration time, by stacking fewer subs, you should get the same result. Just keep in mind, the difference in exposure length here is probably going to be around 3x. So, if you use 5 minute subs at Gain 100, you would probably be using 15 minute subs at Gain 0. For narrow band, you might be using 10-15 minute subs at Gain 100, which would mean 30-45 minute subs at Gain 0.


Edited by Jon Rista, 29 April 2020 - 06:42 PM.

  • Jim Waters, durak, R Botero and 6 others like this

#54 takbiker

takbiker

    Lift Off

  • *****
  • Posts: 12
  • Joined: 13 Oct 2016

Posted 29 April 2020 - 04:02 PM

Hi guys!  This will be my first post here.  I am not a rookie, but I don't post often.  I just received my ASI2600 today.  I ordered direct from ZWO since they briefly had them in stock.  I have been imaging for 15 years or so and transitioned from SBIG mono cameras to a QSI 683 mono camera.  But I also use modified DSLRs such as my T3i.  I decided to try a OSC cooled camera now that the 2600 has such great specs.  I appreciate all the discussions about gain settings, and I plan to start building my darks library over the next few days (raining in Ohio).  I will probably do one set at 0 gain and another at 100 gain (using Maxim DL), with both sets at -10C since we are coming into the warmer months.  I also recently purchased the Triad Ultra filter, and hope this will be a good combination with the 2600, especially with my Hyperstar C11 Edge.  I will let you know how this works out.

 

Tom


  • R Botero likes this

#55 kingjamez

kingjamez

    Vanguard

  • *****
  • Posts: 2,310
  • Joined: 03 Oct 2006
  • Loc: Fairfax, VA

Posted 29 April 2020 - 10:04 PM

I have not had good luck with the triad and hyperstar / rasa scopes. The frequency shift due to the fast light cone reduces aperture and thus focal ratio. The difference between mono with a dedicated fast filter and OSC with the triad is incredible. The triad is great, fantastic even, with slower scopes but just not well suited to hyperstar.

#56 Andy Lucy

Andy Lucy

    Sputnik

  • -----
  • Posts: 49
  • Joined: 09 Apr 2019
  • Loc: East Yorkshire

Posted 01 May 2020 - 07:59 PM

So, for Gain 0, you would have:

 

DRstp0 = log(50109/3.469) * 20 / 6 = 13.87 stops

 

For Gain 100, you would have:

 

DRstp100 = log(16274/1.483) * 20 / 6 = 13.46 stops

 

As you can see, the dynamic range of both gain settings is pretty darn close, both are well over 13 stops, which is EXCELLENT.

...

...If you are ok with shorter exposures, you can get the same results, with the same total integration time, by stacking more subs, at Gain 100. 

The dynamic ranges calculated here are for single subs.

 

Surely it is of more interest to look at the comparative dynamic range of the final stacks, using the two gain settings, each having equal integration time.  ZWO figures give read noise of 1.45 electrons at gain 100 and 3.4 electrons at gain 0.  Assuming that in both cases you aim to just swamp the read noise:  because of the square term you will need to expose 5.5 times longer per sub at gain 0.  Put it another way, for equal integration time you will get 5.5x as many subs at gain 100 as at gain 0.  Stacking 5.5x as many subs gives you an improvement of x2.3 in the noise, which is equivalent to 1.2 extra stops of dynamic range.  

 

Since the dynamic range of a sub at gain 0 is just 0.3 stops better than that of a sub at gain 100, the final stacked image at gain 100 will have a DR improvement of 0.9 stops over the gain 0 final stacked image.

 

Andy


  • nyda83 likes this

#57 mewmartigan

mewmartigan

    Apollo

  • *****
  • Posts: 1,316
  • Joined: 02 Jul 2008
  • Loc: outside of Pittsburgh, PA

Posted 01 May 2020 - 08:08 PM

So for the same integration time gain 100 is better in the end?

#58 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 24,778
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 01 May 2020 - 08:13 PM

The dynamic ranges calculated here are for single subs.

 

Surely it is of more interest to look at the comparative dynamic range of the final stacks, using the two gain settings, each having equal integration time.  ZWO figures give read noise of 1.45 electrons at gain 100 and 3.4 electrons at gain 0.  Assuming that in both cases you aim to just swamp the read noise:  because of the square term you will need to expose 5.5 times longer per sub at gain 0.  Put it another way, for equal integration time you will get 5.5x as many subs at gain 100 as at gain 0.  Stacking 5.5x as many subs gives you an improvement of x2.3 in the noise, which is equivalent to 1.2 extra stops of dynamic range.  

 

Since the dynamic range of a sub at gain 0 is just 0.3 stops better than that of a sub at gain 100, the final stacked image at gain 100 will have a DR improvement of 0.9 stops over the gain 0 final stacked image.

 

Andy

Well, determining the effective dynamic range is a little tougher. You need to account for a few things. Technically speaking, I would call what you have in the stack a change in bit depth, not really a change in dynamic range. I prefer to talk about dynamic range as a physical hardware trait, whereas what you gain with stacking is a matter of bits. 

 

To determine the impact on effective bit depth, you really need to account for the offsets and all the noise, not just read noise. It's a stickier situation at that point, and harder to determine how things really come out in the end...because you can always stack more, vary exposure length, etc. in order to optimize your results. True hardware dynamic range is a relatively simple and fairly fixed thing, whereas effective bit depth depends on exactly how you use the hardware...how you use that physical dynamic range...


Edited by Jon Rista, 01 May 2020 - 08:16 PM.


#59 nyda83

nyda83

    Lift Off

  • -----
  • Posts: 22
  • Joined: 30 Jul 2018
  • Loc: Brescia, Italia

Posted 02 May 2020 - 12:56 AM

The dynamic ranges calculated here are for single subs.

 

Surely it is of more interest to look at the comparative dynamic range of the final stacks, using the two gain settings, each having equal integration time.  ZWO figures give read noise of 1.45 electrons at gain 100 and 3.4 electrons at gain 0.  Assuming that in both cases you aim to just swamp the read noise:  because of the square term you will need to expose 5.5 times longer per sub at gain 0.  Put it another way, for equal integration time you will get 5.5x as many subs at gain 100 as at gain 0.  Stacking 5.5x as many subs gives you an improvement of x2.3 in the noise, which is equivalent to 1.2 extra stops of dynamic range.  

 

Since the dynamic range of a sub at gain 0 is just 0.3 stops better than that of a sub at gain 100, the final stacked image at gain 100 will have a DR improvement of 0.9 stops over the gain 0 final stacked image.

 

Andy

The sharpcap brain give the same results. The use of the gain 100 and offset 15 at the end of the integration give the better dr. I don't know if it's true, after a week that I own the camere I haven't taken an image yet, a week of clouds obviously...



#60 Andy Lucy

Andy Lucy

    Sputnik

  • -----
  • Posts: 49
  • Joined: 09 Apr 2019
  • Loc: East Yorkshire

Posted 02 May 2020 - 05:58 AM

First, I made a mistake in my previous post when calculating the different times needed to achieve maximum dynamic range at gain 0 and gain 100.  The Full Well Capacity is about 3x higher at gain 0 than at gain 100, so it follows that exposure length at gain 0 should be three times longer than at gain 100.  This fits with Jon’s comment in post 53.

For a given integration time you can thus get 3x as many frames at gain 100 as at gain 0.  This corresponds to an improvement of 0.7 stops in what I called dynamic range of the final stack and Jon called effective bit depth.  This still indicates that it would be better overall to operate at gain 100 rather than gain 0.

 

The discussion on the Sharpcap brain referenced by nyda83 is interesting.  The Brain calculates how to get the best final image when you stack all frames taken in a set period of time.  It says in the help file “ ‘Max Dynamic Range’ finds the gain where the final stacked image will have the maximum ratio between the brightest thing that is not quite saturated and the noise level. Max Dynamic Range will often (but not always) choose the minimum gain value.”

I expect that dual gain cameras generally achieve their max dynamic range in the final stack at the gain where the extra amplification kicks in and not at gain 0.

Andy



#61 Umasscrew39

Umasscrew39

    Apollo

  • *****
  • Moderators
  • Posts: 1,347
  • Joined: 07 Dec 2016
  • Loc: Central Florida

Posted 02 May 2020 - 07:03 AM

As a research scientist, it is sometimes easy to get catch up in the data and fall into the old trap of, paralysis by analysis.  So, like with the baby brother ASI533 that I have, I tried my new ASI2600 last night under a very clear Bortle 6 sky with a very bright moon at 63% illumination.  I have had great success with the 533 using a gain of 200 and offset of 10.  Using these same values, I captured just 45 minutes of the Whale Galaxy (180s guided subs) using an Optolong L-Pro filter on a C11" EdgeHD @f/7.  No darks or flats, captures were done with SharpCap and the individual frames were aligned, integrated and post-processed using PixInsight.  I may take the time to do the SC sensor analysis (or try 100 gain) but I was impressed by the low noise and excellent color of the image after just 45 minutes under my skies using these settings - very similar to the 533.  I think if I take the time to do this for a few hours and add both flats and darks, it can become an excellent image.     

Attached Thumbnails

  • Whale Galaxy2.jpg

  • mewmartigan likes this

#62 buckeyestargazer

buckeyestargazer

    Vendor - Buckeyestargazer.net

  • *****
  • Vendors
  • topic starter
  • Posts: 5,395
  • Joined: 12 Jan 2008
  • Loc: IN, USA

Posted 02 May 2020 - 08:34 AM

All of these numbers and analysis is great, but I think the real world result will be that I will see very little difference, if any at all, between 50 x 10min subs at gain 0 and 166 x 3min subs at gain 100.  And I'd rather have 50 (51mb) files on my hard drive than 166 (51mb) files on my hard drive.  That's just a personal choice for my needs and sky conditions.   


  • R Botero and gatsbyiv like this

#63 Umasscrew39

Umasscrew39

    Apollo

  • *****
  • Moderators
  • Posts: 1,347
  • Joined: 07 Dec 2016
  • Loc: Central Florida

Posted 06 May 2020 - 05:15 AM

All of these numbers and analysis is great, but I think the real world result will be that I will see very little difference, if any at all, between 50 x 10min subs at gain 0 and 166 x 3min subs at gain 100.  And I'd rather have 50 (51mb) files on my hard drive than 166 (51mb) files on my hard drive.  That's just a personal choice for my needs and sky conditions.   

Completely agree.  That was exactly my point in the post above yours and this same conclusion became evident on a very similar thread discussing the ASI533MC Pro.  


  • buckeyestargazer and R Botero like this

#64 evan9162

evan9162

    Surveyor 1

  • *****
  • Posts: 1,714
  • Joined: 20 Sep 2013
  • Loc: ID

Posted 06 May 2020 - 09:26 AM

There are advantages to shorter/more subs:

 

-fewer rejected subs due to guiding errors, satellite/airplane trails, seeing, focus shift, flexure, etc. 

 

-larger number of subs will be more effective for pixel rejection algorithms

 

-a little bit of a resolution/sharpness benefit with shorter subs (less seeing blur)

 

-You can effectively drizzle your integration for better color accuracy (vs. debayering) and a little resolution bump

 

-Higher gain reduces quantization error




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics