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

Looking at the histogram and counting max pixels is not enough

  • Please log in to reply
9 replies to this topic

#1 unimatrix0

unimatrix0

    Fly Me to the Moon

  • *****
  • topic starter
  • Posts: 6,796
  • Joined: 03 Jan 2021

Posted 03 October 2023 - 01:08 PM

Ok, so this is just my personal experiment, hoping to help out some people who just can't get a good enough image, regardless how good they guide or what post processing they do. 

 

If your image ends up noisy, even after plenty of subs, read on. 

 

First thing first, to help yourself just get Sharpcap Pro and familiarize yourself with the software. I started out my astro-journey using Sharpcap Pro, after I spent time doing plenty of visual observing only. 

My first stop before I image (unless I use the Asiair) is still Sharpcap Pro- doing it for first time focusing, sky brightness measurements and other stuff. 

 

So, the first thing I always do when I get a new camera is do a sensor analysis.  It can take some time and some practice to get the right amount of light and so on, but it's so worth it. 

 

Here is what I'm taking about, related to the title: 

 

I see people just copying what others do, regardless of which camera or what their sky conditions or what filter is used. While that could be a good starting point, but that won't guarantee that you gonna end up with a good image regardless. 

 

Let me show you a demonstration with my QHY183C camera.  I'm doing this during daytime so the exposures are tiny, but it's the same thing at night with longer exposures. 

 

The general consesus for this camera is "use gain 11 and offset 30"-   Yeah ok.. except the fact that the 183C sensor doesn't behave like the 533 or the 2600. While it has a "unity" gain, that doesn't matter, because unlike the 533 or 2600, there is no low gain conversion versus high gain conversion stuff happening, there is no "step" in the analysis, the camera doesn't gain back dynamic range, neither the noise have any sort of "step-down" , but it's a gradual decrease.  

https://www.qhyccd.c...-camera-qhy183/

Ok, so regardless, see the picture below what the camera shows "when the histogram is a bit to the right and the histogram bump is a nice line".  

Oh cool, right? 

 

I'm not over exposing and my histogram is not "hanging off" the left edge, must be good settings!   NOPE. 

 

med_gallery_355785_18054_92546.jpg

 

The correct would be this one, but let me get to it. 

Just for a demo, the Green bar on top is the "good" signal to noise ratio area, the red and orange are the not so good area. It's dominated by more noise than signal. Your histogram bump should be past the red and orange area. 

med_gallery_355785_18054_14850.jpg

 

 

 

Look carefully what the smart histogram is showing you. 

I made 2 more screen captures about the red and the orange area of the histogram what it represents.  Please ignore the bottom green/light green bar, that's for the 8Bit information only, not the 16,  we don't care about that right now. 

Read the the pop up note from Sharpcap:   That's the red area. 
gallery_355785_18054_27169.jpg

 

Orange Area: 

gallery_355785_18054_2484.jpg

 

So basically having your histogram bump in either of those areas would mean, that you are capturing more noise than signal. You're wasting a lot of your time and pumping out subs that have the noise dominate over the signal and you can stack the heck out of them, you will get an image, but you could have gotten a better image if you would have followed the recommendation of the software. 

 

Let's say I'm a free thinker and don't care about what anything tells me and instead of increasing the exposure as recommended, I increase the offset. 

Look what happens. 

 

med_gallery_355785_18054_61456.jpg

 

Did that fix the problem? 

NO. 

I just increased the area (cutting down further on the dynamic range) that will be dominated by noise mostly.  The histogram bump moved to the right and without seeing this smart histogram, I would assume I fixed my bottom noise issues, but I did not, I just reduced the capability of the camera even more. My histogram bump is still sitting in the orange zone (which is dominated by noise and not signal). 

 

 

I'm gonna be a free thinker again, and this time I'm gonna increase the gain!  That will fix it! 

Here is what happens, after bumping up the gain: 

med_gallery_355785_18054_121073.jpg

 

Dit that fix my noise dominated image? 

Nope!  The majority of the signal (mostly the faint) is still sitting in the red and orange, the gain only amplified the little signal and majority of the noise- hence the gain is called amplification, but does not increase photon input) 

 

So, let's listen to Sharpcap pro, and instead of messing with the gain and the offset, let's do a sky analysis. Again, I'm doing this during day time and the exposures are very short, but it's the same deal at night, except the exposures gonna be longer and the recommendations will be those longer exposures .  So, it's just the matter of scale and brightness, but it's the same thing. 

 

I do the sharpcap sky analysis (my garage wall inside), and I want to go for max dynamic range.  I want low noise, so I chose 1%.   (5% or 10% would be more realistic for this camera, but for the sake of demo I chose 1) 

 

Note, that it recommends me to drop the offset to 12,  drop the gain to 0 and increase the exposure.  This will vary for everyone, these numbers (again daytime here) will be different for everyone and every camera. Even cameras of the same model and brand will vary to a certain number and only after local - at your home - sensor analysis can tell you the right numbers at that particular sky brightness you happen to have - wherever you are imaging.   

(Black level = Offset)

 

med_gallery_355785_18054_192188.jpg

 

gallery_355785_18054_92216.jpg

 

 

See what happens when I apply the settings, 

Take a note of the histogram bump(s), since I'm using a colored camera. Note that they moved out of the red and orange zone, the majority of signal is outside the camera noise level. This is the optimal gain and offset and exposure length to capture the DSO I'm going after (in this case, my garage wall inside) 

 

 

 

Now the faint stuff is no longer buried in the noise, the dynamic range will be the max without clipping the darks.  

BTW, I stopped following this idea that either NINA or whatever else telling me how many stars are over exposed.  NINA shows a unlinked stretched image and the histogram is showing of an average of 3 colors if it's an OSC camera. While it's cool to have it, it's not telling you the whole story. 

And if you got a camera with an amp glow, then forget about counting your overexposed pixels, because the amp glow is also calculated in and will mislead you for sure,  whether you are under or over exposing.  Stars peaking through an unstretched image is still fine, as long ast they are super bright to begin with. 

BTW, modern cameras like the 533 and 2600 are much much better when it comes to signal to noise ratio so yes, you can do a lot shorter exposures than needed, but there is a limit there too how short.  

med_gallery_355785_18054_134393.jpg

 

 

If you disagree with everything seen here, please contact the person behind Sharpcap Pro-  Dr. Robin Glover-  over the sharpcap forums and tell him he is wrong. grin.gif


Edited by unimatrix0, 03 October 2023 - 01:27 PM.

  • Pingu and Desertanimal like this

#2 dciobota

dciobota

    Soyuz

  • *****
  • Posts: 3,735
  • Joined: 07 Aug 2007
  • Loc: No longer on this site in protest of poor site moderation

Posted 03 October 2023 - 01:14 PM

I've always preached to ignore overexposed pixel counts to some extent and concentrate on bringing not the peak, but rather the faint signal of the object above the noise.
The only thing is that it's sometimes hard to tell where that faint signal is, especially if you're in a high lp zone.

#3 unimatrix0

unimatrix0

    Fly Me to the Moon

  • *****
  • topic starter
  • Posts: 6,796
  • Joined: 03 Jan 2021

Posted 03 October 2023 - 01:57 PM

I've always preached to ignore overexposed pixel counts to some extent and concentrate on bringing not the peak, but rather the faint signal of the object above the noise.
The only thing is that it's sometimes hard to tell where that faint signal is, especially if you're in a high lp zone.

It really comes down to the camera what it's capable really, when it comes to "expose long enough to get the faint signal, but don't over expose the stars" .   Thats when a large -e well depth helps. 

 

 The sky analysis will take the high light pollution in account and will still recommend the best settings that won't let the images dominated by noise.  Actually it's harder to get a noisy image with a lot of light pollution, but the inherent sensor noise is still there that needs to be get rid of. 

 

I did an experiment last night, I exposed the 183C for 5 minutes on gain 0 on Andromeda. The 183 is known to have a small well depth. (533 , 2600 a lot more)   The recommended (By sharpcap pro) was  gain0, offset 12 and 230 seconds per sub if I only wanted 1% noise.  That sounds awfully lot, but it's really not (at gain 0). and I'm at Bortle 5.  If I would have chosen to settle with 5% or 10% noise, the recommended exposure was like between 2-3 minutes per sub. Again, sounds too much and I did do a few subs and some stars did peak through the unstretched image. Regardless the image looked good while stretched, the yellow stars were yellow, blue stars still blue, I didn't notice anything "blown". 

 

Instead I did the 5 minutes and I examined the stars.  I was concentrating on the colors, so it happens that the yellow stars were still yellow so I assumed I didn't blow the stars.  But I did, but only the bright blue/white ones.  The majority of the stars like 95% were not over exposed at all. The core of the Galaxy was over exposed, but again, if someone wants to capture Andromeda will most likely blow the cores with no filters and of course with 5 minute subs, which is an extra 1.5 minute over the recommended. 

 

 

 

As far as I'm noticed, I find it very hard to over expose with the 533 and any narrowband filter at my bortle 5 sky.  When it comes to broadband, it can be a different story, but still I've been using a lot shorter exposures than recommended. 


Edited by unimatrix0, 03 October 2023 - 02:00 PM.


#4 Andros246

Andros246

    Apollo

  • -----
  • Posts: 1,091
  • Joined: 24 Oct 2022

Posted 03 October 2023 - 02:17 PM

Just to add I think that even if you do everything "right" your still at the mercy of the limit of your skies which under allot of light pollution you wont be able to capture those feint details anyway unless you have an outrageous amount of exposure time and i mean an outrageous amount.

 

To newbies theres's no magic exposure length for getting great results head for darker skies.


Edited by Andros246, 03 October 2023 - 02:21 PM.


#5 RichNH

RichNH

    Viking 1

  • -----
  • Posts: 865
  • Joined: 06 May 2005
  • Loc: Brookline, New Hampshire, USA

Posted 03 October 2023 - 03:04 PM

I took a pic of M31 last night, first time ever using the ASI533MC camera (which is my first camera), so I know nothing.  I did 30 second exposures with the gain set at 200 and the temp set at -19C.  I took 100 subs.  I'm planning on taking more subs tonight, do you think I should keep the same settings as last night or should I change them?  I'm using ASIStudio to do all this work.  All I did today was finish taking the flats this morning and then stacked everything.  This is what ASIDeepStack gave me for an image.  I do have the final, stacked fits file as well as all the original fits files.  I plan on using these and others to learn how to post-process images.

 

Attached Thumbnails

  • 231002_M31.jpg

  • acrh2 likes this

#6 KTAZ

KTAZ

    Skylab

  • *****
  • Posts: 4,424
  • Joined: 09 Apr 2020
  • Loc: Scottsdale, AZ

Posted 03 October 2023 - 03:10 PM

I took a pic of M31 last night, first time ever using the ASI533MC camera (which is my first camera), so I know nothing.  I did 30 second exposures with the gain set at 200 and the temp set at -19C.  I took 100 subs.  I'm planning on taking more subs tonight, do you think I should keep the same settings as last night or should I change them?  I'm using ASIStudio to do all this work.  All I did today was finish taking the flats this morning and then stacked everything.  This is what ASIDeepStack gave me for an image.  I do have the final, stacked fits file as well as all the original fits files.  I plan on using these and others to learn how to post-process images.

 

The point to make is that there is no single answer. Asking folks here whether you should change your settings is impossible to answer. Many, many factors contribute to the proper settings for exposure. That is why the OP is using Sharpcap Pro as a baseline for analysis. 

 

As a beginner, you should focus on understanding what effects changing exposure settings have on your subs. Sharpcap Pro is a great tool to start to learn the ropes, and will do the work for you until you learn more. It is dirt cheap for what it provides.



#7 RichNH

RichNH

    Viking 1

  • -----
  • Posts: 865
  • Joined: 06 May 2005
  • Loc: Brookline, New Hampshire, USA

Posted 03 October 2023 - 03:11 PM

OK, I just figured that by looking at the image someone might have observations such as too much gain or whatever.  Looking at the SharpCap website it apparently has a trial version so I've installed that and will give it a try tonight (2 nights in a row with clear skies, woohoo!).


Edited by RichNH, 03 October 2023 - 03:37 PM.

  • KTAZ likes this

#8 acrh2

acrh2

    Mercury-Atlas

  • *****
  • Posts: 2,821
  • Joined: 16 Mar 2021

Posted 03 October 2023 - 03:58 PM

I took a pic of M31 last night, first time ever using the ASI533MC camera (which is my first camera), so I know nothing.  I did 30 second exposures with the gain set at 200 and the temp set at -19C.  I took 100 subs.  I'm planning on taking more subs tonight, do you think I should keep the same settings as last night or should I change them?  I'm using ASIStudio to do all this work.  All I did today was finish taking the flats this morning and then stacked everything.  This is what ASIDeepStack gave me for an image.  I do have the final, stacked fits file as well as all the original fits files.  I plan on using these and others to learn how to post-process images.

 

According to Dr. Glover,

 

https://www.youtube....h?v=3RH93UvP358

http://www.tools.sharpcap.co.uk/

 

given your light pollution (Bortle 3.5 at your location), your telescope speed (f/7), and your camera, you should expose longer - 2-3 minutes.

 

 

You will blowing out some brighter stars, so probably stick to 1-2 minutes, depending on the target.

 

 

According to Dr. acrh2,

 

you don't need to cool your camera to -20C. You could cool it to 0C and that would still be fine. Unless you wanted to keep two sets of darks, one for summer and one for winter, I would recommend -5C. You climate is slightly milder than my own, and I've been good with -5C. Except that I don't image when it gets below -8-10C, cause it's just too **** cold to set up the mount. 

The last week was all 80F at the start of imaging sessions, so it would be impossible to cool the sensor to -19C. Certainly, it would be impossible in the summer.

 

There's no real reason to use anything other than 100 gain with this camera, unless you have a really fast and large telescope (then you would use 0.)

 

You have 9 hours of dark now, use it.

 

So, this doctor recommends - 

consider raising the temp to -5C,

raise your exposure time to 2 minutes, lower your gain to 100,

try and get more than 50 minutes of data. 


  • RichNH, dciobota and Spaceman 56 like this

#9 dciobota

dciobota

    Soyuz

  • *****
  • Posts: 3,735
  • Joined: 07 Aug 2007
  • Loc: No longer on this site in protest of poor site moderation

Posted 03 October 2023 - 04:16 PM

+1 on what he said above. Note that gain 0 and gain 100 are the only two gains that give you the best dynamic range. While there's argument between 0 and 100 as to which is best, I tried both and settled on 100. I also now shoot at 0C which saves on battery power and shows no noticeable increase in noise from -10C that I was using before.
  • RichNH and KTAZ like this

#10 KTAZ

KTAZ

    Skylab

  • *****
  • Posts: 4,424
  • Joined: 09 Apr 2020
  • Loc: Scottsdale, AZ

Posted 03 October 2023 - 07:58 PM

I live in the Arizona desert, so I run 0 in summer and -10 in winter (and maintain 2 sets of darks). There is really no need to go any lower.


  • RichNH and dciobota like this


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