Page 8 of 8 FirstFirst ... 678
Results 71 to 75 of 75
Like Tree20Likes

Thread: Open Image Processing 101

  1. #71
    Phil Leigh's Avatar
    Phil Leigh is offline HYPER GIANT
    Points: 34,455, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Overall activity: 0%
    Achievements:
    5 Threads Achievement!50 Posts Achievement!200+ Posts Achievement!First 1000 Experience Points400+ Posts Achievement
    Join Date
    Nov 2011
    Location
    Cryers Hill, High Wycombe, UK
    Posts
    5,867
    Points
    34,455
    Level
    100
    Thanks
    451
    Thanked 2,018x 1,651 Posts
    Blog Entries
    8

    Default Re: Open Image Processing 101



    Quote Originally Posted by s24man View Post
    I'm not talking about the histogram--I understand that the histogram is just showing what's on the 8 bit .jpg. What I'm talking about is image statistics. Where this whole thing started was that Michael is trying to measure his flats. Many programs will give you min, max and mean ADU for the image (which is independent of the histogram stretch). Nebulosity, Pixinsight, SGP, Maxim, etc. all do it. If BEOS had that also, then it wouldn't matter what the histogram showed. I underexposed my flats for 2 years before I realized that the histogram was useless in BEOS.

    And neither was I... I was saying that what would be very useful would be something simple based on (say) the mean value of the RAW data vs the half-maximum for the specific camera bit-depth... so for example; if you had a 14-bit camera saturation occurs at somewhere below 14-bits (16384) - so a good target to aim for flats would be 4096-8192 as the mean (although in practice anything above 1024 works fine...DSLR's are very linear). If BYE could say "Increase exposure to x (seconds) to optimise flat" - that would be very cool and easy for people to understand.

    Alternatively it could be based on a "too low" warning... like if the flat mean was below 1024 i could say "Increase exposure!" or something.

    Min/Max/Mean ADU is useful for CCD imagers. ADU is not a concept that has much meaning to most DSLR users... especially as there are both 12 and 14 bit cameras in play.

    I'm not arguing AGAINST have image stats...just that maybe something else (something simpler) would be useful too.
    Telescope: SWEquinox ED80 Pro (Schott/Ohara Fluorite)+Baader Steeltrack, C8 XLT Mount: Skywatcher HEQ5 Pro, Rowan belt drive on Pillar mount+EQDIR+ADM saddle/dovetails Camera: JTW1100D mono , CentralDS 600, ASI120MM. ASI120MC, IDAS & Baader NB 2-inch filters, TS Filter Drawer System Accessories: TV Nagler T4 12 mm, Baader 8-24 MkIII, Ortho 5mm, Aspheric 31mm, Meade 12mm reticle, TV Powermate 2x & 2.5x, JMI+BAST Motorfocus+FCUSB, TV 0.8x FR/FF, Celestron 0.63FR, Orion RACI & Mag Mini, Canon IS 15x50 Bins
    Software: EQMOD/Stellarium/Autostakkert!/Registax/BackyardEOS/PixInsight,PHD2,AlignMaster, AstroTortilla, Straton

  2. #72
    s24man's Avatar
    s24man is offline SUPER GIANT
    Points: 12,059, Level: 76
    Level completed: 3%, Points required for next Level: 341
    Overall activity: 0%
    Achievements:
    2 Posts Achievement20 Posts Achievement!50 Posts Achievement!5 Threads Achievement!200+ Posts Achievement!
    Join Date
    Mar 2014
    Location
    San Francisco Bay Area
    Posts
    1,466
    Points
    12,059
    Level
    76
    Thanks
    90
    Thanked 425x 373 Posts

    Default Re: Open Image Processing 101

    Completely agree with everything Phil wrote, with one quibble. The DSLR cameras have a rather high bias level of about 2048. Having a flat that's only 1000 over the bias may not help, since it's on the very low end. Better err on the side of higher exposures. Basically you want to fill about 45-50% of the full well depth as you're well within the linear response of the camera. But it can be confusing if you're comparing Digic 2 12 bit cameras and Digic 3 14 bit cameras as the ADU would be different. (I made this mistake helping a guy online with this issue and it was about 2 days before we realized the error). If the program were just to do an internal measurement and say, "Too dark" then that would be fine. Unfortunately, it doesn't, so we're stuck with imaging statistics. Which can be and are confusing to many. 14 bit Canons like Mike has have a max ADU of about 16,000, but when opening up the image in nebulosity it upscales the image to a 16 bit 65,000. So now you are looking for one number, you get it and it's still low, and you don't realize it's upscaling.

    Basically getting correct flat measurments is a bit of a pain in the ass. But really, you only need to do it once as subsequent flats, assuming the light source and camera is constant, should be more or less the same.
    Stuart Forman


    To view links or images in signatures your post count must be 5 or greater. You currently have 0 signatures.
    Engineering Company 140mm f7 apochromat with TEC FF on Astro-Physics Mach 1 with RAPAS. Also with A-P 27tvph
    To view links or images in signatures your post count must be 5 or greater. You currently have 0 signatures.
    . Camera QSI 683ws-8 Astrodon Gen 2 LRGBHaO3S2 (5nm), Starlight Instruments Focus Boss focuser,
    To view links or images in signatures your post count must be 5 or greater. You currently have 0 signatures.
    : SGP, Pixinsight, Guiding with A-P Guiding bracket, 10X60 Vario
    To view links or images in signatures your post count must be 5 or greater. You currently have 0 signatures.
    , Lodestar, PHD. Rob Miller TRI36L tripod on JMI Wheely Bars. Twitter @s24man

  3. #73
    Phil Leigh's Avatar
    Phil Leigh is offline HYPER GIANT
    Points: 34,455, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Overall activity: 0%
    Achievements:
    5 Threads Achievement!50 Posts Achievement!200+ Posts Achievement!First 1000 Experience Points400+ Posts Achievement
    Join Date
    Nov 2011
    Location
    Cryers Hill, High Wycombe, UK
    Posts
    5,867
    Points
    34,455
    Level
    100
    Thanks
    451
    Thanked 2,018x 1,651 Posts
    Blog Entries
    8

    Default Re: Open Image Processing 101

    Quote Originally Posted by s24man View Post
    Completely agree with everything Phil wrote, with one quibble. The DSLR cameras have a rather high bias level of about 2048. Having a flat that's only 1000 over the bias may not help, since it's on the very low end. Better err on the side of higher exposures. Basically you want to fill about 45-50% of the full well depth as you're well within the linear response of the camera. But it can be confusing if you're comparing Digic 2 12 bit cameras and Digic 3 14 bit cameras as the ADU would be different. (I made this mistake helping a guy online with this issue and it was about 2 days before we realized the error). If the program were just to do an internal measurement and say, "Too dark" then that would be fine. Unfortunately, it doesn't, so we're stuck with imaging statistics. Which can be and are confusing to many. 14 bit Canons like Mike has have a max ADU of about 16,000, but when opening up the image in nebulosity it upscales the image to a 16 bit 65,000. So now you are looking for one number, you get it and it's still low, and you don't realize it's upscaling.

    Basically getting correct flat measurments is a bit of a pain in the ass. But really, you only need to do it once as subsequent flats, assuming the light source and camera is constant, should be more or less the same.
    That's a 2048 bias on the 16-bit scale...so 512 for a 14-bit camera? (and 128 for a 12-bit one)

    Modern DSLR's CMOS sensor linear response extends from 2x bias level to 1 bit below maximum (so from 1024 to 8192 for a 14-bit camera in 14-bit terms or 256-2048 for a 12-bit camera in 12-bit terms). You don't actually need to get to half full well as per CCD.


    You're right it is extremely easy to get confused! I muddled up my numbers.

    For a 14-bit camera, lowest flat that works would be > 1024/4096 (14/16 bit scales) and the highest would be 8192/32768
    For a 12-bit camera it would be >256/4096 (12/16 bit scales) and 2048/32768

    This assumes that the absolute bias level for 12 and 14-bit cameras is the same...? All of my cameras are 14-bit so I can't easily check.
    Telescope: SWEquinox ED80 Pro (Schott/Ohara Fluorite)+Baader Steeltrack, C8 XLT Mount: Skywatcher HEQ5 Pro, Rowan belt drive on Pillar mount+EQDIR+ADM saddle/dovetails Camera: JTW1100D mono , CentralDS 600, ASI120MM. ASI120MC, IDAS & Baader NB 2-inch filters, TS Filter Drawer System Accessories: TV Nagler T4 12 mm, Baader 8-24 MkIII, Ortho 5mm, Aspheric 31mm, Meade 12mm reticle, TV Powermate 2x & 2.5x, JMI+BAST Motorfocus+FCUSB, TV 0.8x FR/FF, Celestron 0.63FR, Orion RACI & Mag Mini, Canon IS 15x50 Bins
    Software: EQMOD/Stellarium/Autostakkert!/Registax/BackyardEOS/PixInsight,PHD2,AlignMaster, AstroTortilla, Straton

  4. #74
    s24man's Avatar
    s24man is offline SUPER GIANT
    Points: 12,059, Level: 76
    Level completed: 3%, Points required for next Level: 341
    Overall activity: 0%
    Achievements:
    2 Posts Achievement20 Posts Achievement!50 Posts Achievement!5 Threads Achievement!200+ Posts Achievement!
    Join Date
    Mar 2014
    Location
    San Francisco Bay Area
    Posts
    1,466
    Points
    12,059
    Level
    76
    Thanks
    90
    Thanked 425x 373 Posts

    Default Re: Open Image Processing 101

    Hi Phil,
    Okay I was curious, so I went and measured some light and bias subs from my old DSLR, my QSI, and Mike's subs. I did this with CCD stack.

    I had a Rosette from my DSLR and one from my QSI. I looked at one of the bright stars in the DSLR (a 14 bit Digic 3 processor Canon T1i) and the ADU was about 15,000. The same star on the QSI was 65,000. So they were both saturated, and the 14 bit was not scaled by CCD stack

    Then I went to a bias. On my QSI, a couple of different bias at -20C was about 250. -30 was the same (as you would expect)

    Then I measured a couple of different biases from Canon T1i. They were both 1025.

    Then I measured Mike's mean bias from his 60Da which is the newer Digic 4 processor. I measured two bias frames and they were both 2049. So the base bias ADU is about 1020 higher than the Digic 3. But both are 14 bit, and I established at the beginning of this post already that CCDstack is not scaling like, say, Nebulosity.

    So, if his base count is 2049, and his high is 15,000, then that's a difference of about 13,000. Half that is 6500. 6500 + 2050 = 8550. So that's more or less what I would shoot for when shooting flats. But he's using Nebulosity to measure which is scaled. So doing some cocktail napkin calculations, 8550 is equivalent to about 34,000 scaled up to 16 bits, so that would be what I would shoot for. Again, more or less.
    Stuart Forman


    To view links or images in signatures your post count must be 5 or greater. You currently have 0 signatures.
    Engineering Company 140mm f7 apochromat with TEC FF on Astro-Physics Mach 1 with RAPAS. Also with A-P 27tvph
    To view links or images in signatures your post count must be 5 or greater. You currently have 0 signatures.
    . Camera QSI 683ws-8 Astrodon Gen 2 LRGBHaO3S2 (5nm), Starlight Instruments Focus Boss focuser,
    To view links or images in signatures your post count must be 5 or greater. You currently have 0 signatures.
    : SGP, Pixinsight, Guiding with A-P Guiding bracket, 10X60 Vario
    To view links or images in signatures your post count must be 5 or greater. You currently have 0 signatures.
    , Lodestar, PHD. Rob Miller TRI36L tripod on JMI Wheely Bars. Twitter @s24man

  5. The Following User Says Thank You to s24man For This Useful Post:

    Mike_Lewis (04-03-2014)

  6. #75
    Phil Leigh's Avatar
    Phil Leigh is offline HYPER GIANT
    Points: 34,455, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Overall activity: 0%
    Achievements:
    5 Threads Achievement!50 Posts Achievement!200+ Posts Achievement!First 1000 Experience Points400+ Posts Achievement
    Join Date
    Nov 2011
    Location
    Cryers Hill, High Wycombe, UK
    Posts
    5,867
    Points
    34,455
    Level
    100
    Thanks
    451
    Thanked 2,018x 1,651 Posts
    Blog Entries
    8

    Default Re: Open Image Processing 101

    Thanks... by coincidence I'd just discovered that some older 14-bit camera models have the 1024 bias level.

    As far as DSLR flats go the actual level is not particularly critical provided it is at least more than twice the bias level - the linear response range for the CMOS sensors is very wide.
    Telescope: SWEquinox ED80 Pro (Schott/Ohara Fluorite)+Baader Steeltrack, C8 XLT Mount: Skywatcher HEQ5 Pro, Rowan belt drive on Pillar mount+EQDIR+ADM saddle/dovetails Camera: JTW1100D mono , CentralDS 600, ASI120MM. ASI120MC, IDAS & Baader NB 2-inch filters, TS Filter Drawer System Accessories: TV Nagler T4 12 mm, Baader 8-24 MkIII, Ortho 5mm, Aspheric 31mm, Meade 12mm reticle, TV Powermate 2x & 2.5x, JMI+BAST Motorfocus+FCUSB, TV 0.8x FR/FF, Celestron 0.63FR, Orion RACI & Mag Mini, Canon IS 15x50 Bins
    Software: EQMOD/Stellarium/Autostakkert!/Registax/BackyardEOS/PixInsight,PHD2,AlignMaster, AstroTortilla, Straton

 

 
Page 8 of 8 FirstFirst ... 678

Similar Threads

  1. Help processing M20 image
    By akulapanam in forum Astrophotography Image Processing Forum
    Replies: 8
    Last Post: 08-08-2012, 03:18 AM
  2. Open Image Processing Template and Information for Posting Your Image for Open Processing by the Community
    By admin in forum Astrophotography Image Processing Forum
    Replies: 1
    Last Post: 02-29-2012, 05:47 PM
  3. Need help with image processing
    By dragracingdan in forum Astrophotography Image Processing Forum
    Replies: 3
    Last Post: 02-08-2012, 08:12 PM
  4. Need help on image processing
    By Xavier in forum UK Astronomy Forum
    Replies: 2
    Last Post: 07-08-2004, 06:27 AM
  5. Need help with image processing
    By Jeff Polston in forum Amateur Astronomy Forum
    Replies: 3
    Last Post: 03-02-2004, 08:23 PM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  
Powered by vBulletin® Version 4.2.0
Powered by vBulletin®
All times are GMT. The time now is 11:35 AM.