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

Morphing Flats?

  • Please log in to reply
18 replies to this topic

#1 Dynan

Dynan

    Mercury-Atlas

  • *****
  • topic starter
  • Posts: 2704
  • Joined: 11 Mar 2018
  • Loc: NOLA

Posted 19 April 2019 - 08:26 PM

This thread ties into this thread discussing proper exposure of flats:

 

https://www.cloudyni...lats-half-flat/

 

So...I determined that, as described in post #12 (https://www.cloudyni...flat/?p=9267915) I would expose my flats to get ≈0.125 Mean Result.

 

Here is a screenshot of the PI Image Statistics of one of the flats as it was being taken:

 

2019_0409_FLATS_PI_STATS_1S.jpg

 

The histogram shown in BYEOS looked something like I thought it should (R and B at 50%, with G somewhat higher since it has two pixels per R/B):

 

2019_0409_FLATS_BYEOS_HISTO_S.jpg

 

BUT when I open the flats in PI now, I get this:

 

2019_0409_FLATS_PI_STATS_2S.jpg

 

Can someone please explain where 60% of the value went?

This is the SECOND set that acts the exact same way.

I believe a .05 Mean Flat isn't good enough for calibration. Am I wrong?

I did no processing to the Flat, just a stretch for this post.

 

Am I doing something wrong in stat analysis in PI? The histogram in BYEOS looks good, I think.?.?? Am I asking the wrong questions?

 

Thanks for any help you can give solving this flat conundrum.



#2 bmhjr

bmhjr

    Apollo

  • ****-
  • Posts: 1185
  • Joined: 02 Oct 2015
  • Loc: Colorado

Posted 19 April 2019 - 08:55 PM

The first flat you show is RGB. Is it a jpeg or debayered? You should be using the raw file.

 

In PI, open the Format Explorer and make sure that RAW is set to Pure Raw.  This will keep the raw format when you open the image in PI to measure the statistics.  What format are you using, .cr2?

 

Workspace01.jpg

 


Edited by bmhjr, 19 April 2019 - 09:21 PM.


#3 Dynan

Dynan

    Mercury-Atlas

  • *****
  • topic starter
  • Posts: 2704
  • Joined: 11 Mar 2018
  • Loc: NOLA

Posted 19 April 2019 - 09:28 PM

Undebayered, RAW (cr2 from Canon).

 

Here's a shot of my Pure Raw setting in PI. I believe it was already set when I open the flat.

 

PI RAW SETTINGS.jpg



#4 bmhjr

bmhjr

    Apollo

  • ****-
  • Posts: 1185
  • Joined: 02 Oct 2015
  • Loc: Colorado

Posted 19 April 2019 - 09:35 PM

Interesting. I know my histogram in BYEOS shows all the way to the right, almost clipped. Also, the max Normalized value in PI is 0.25. So midway is somewhere around .125. It doesnt need to be exact.

#5 Dynan

Dynan

    Mercury-Atlas

  • *****
  • topic starter
  • Posts: 2704
  • Joined: 11 Mar 2018
  • Loc: NOLA

Posted 19 April 2019 - 09:41 PM

And this is a flat (same batch) debayered. Read in Stats as RGB. I still see Low Mean, as you suggested I aim for the .125

 

PI RAW SETTINGS RGB.jpg

 

What am I misunderstanding?



#6 bmhjr

bmhjr

    Apollo

  • ****-
  • Posts: 1185
  • Joined: 02 Oct 2015
  • Loc: Colorado

Posted 19 April 2019 - 09:54 PM

Im not sure what's the issue.  You may need to expose longer.  What you are showing is not exposed long enough.  Here is my histogram in BYEOS.  Can you add one of your flats to dropbox or something?

 

histo.jpg


Edited by bmhjr, 19 April 2019 - 09:58 PM.


#7 Dynan

Dynan

    Mercury-Atlas

  • *****
  • topic starter
  • Posts: 2704
  • Joined: 11 Mar 2018
  • Loc: NOLA

Posted 19 April 2019 - 10:12 PM

Try this:

 

https://www.dropbox....2?role=personal

 

If you can't access it, it's because I'm pretty dumb for a smart guy.



#8 bmhjr

bmhjr

    Apollo

  • ****-
  • Posts: 1185
  • Joined: 02 Oct 2015
  • Loc: Colorado

Posted 19 April 2019 - 10:16 PM

Try this:

 

https://www.dropbox....2?role=personal

 

If you can't access it, it's because I'm pretty dumb for a smart guy.

Link is not working for me, says does not exist



#9 Dynan

Dynan

    Mercury-Atlas

  • *****
  • topic starter
  • Posts: 2704
  • Joined: 11 Mar 2018
  • Loc: NOLA

Posted 19 April 2019 - 10:19 PM

Hence the dumb...

 

Try this please:

 

https://www.dropbox....s911ms.CR2?dl=0



#10 bmhjr

bmhjr

    Apollo

  • ****-
  • Posts: 1185
  • Joined: 02 Oct 2015
  • Loc: Colorado

Posted 19 April 2019 - 10:25 PM

Hence the dumb...

 

Try this please:

 

https://www.dropbox....s911ms.CR2?dl=0

Yes it looks under exposed.  Try doubling your exposure time.



#11 Dynan

Dynan

    Mercury-Atlas

  • *****
  • topic starter
  • Posts: 2704
  • Joined: 11 Mar 2018
  • Loc: NOLA

Posted 19 April 2019 - 10:31 PM

I'll DO that. Thanks BMH. I went on your recommend in the first place. I'll bring those flat exposures up to .250 and see what happens.

 

I DID notice no vignetting in any of my flats, which I assume should be there...along with dust bunnies.



#12 bmhjr

bmhjr

    Apollo

  • ****-
  • Posts: 1185
  • Joined: 02 Oct 2015
  • Loc: Colorado

Posted 19 April 2019 - 10:39 PM

I'll DO that. Thanks BMH. I went on your recommend in the first place. I'll bring those flat exposures up to .250 and see what happens.

 

I DID notice no vignetting in any of my flats, which I assume should be there...along with dust bunnies.

The mean should read close to 0.125 but the max should never go over 0.25 in the way PI displays the stats for a 14 bit CR2 file.  At least for my camera  smile.gif

 

Bill



#13 Dynan

Dynan

    Mercury-Atlas

  • *****
  • topic starter
  • Posts: 2704
  • Joined: 11 Mar 2018
  • Loc: NOLA

Posted 19 April 2019 - 10:53 PM

Now I'm confused. I DID expose to the mean of .125 and I got the results you saw in the Dropbox. If I do the same thing, I'll get the same low results. Something isn't meshing here.

 

Should I use the Normalized Real, or RGB balance for the mean?

 

Am I asking the right questions?



#14 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 23248
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 19 April 2019 - 10:57 PM

Interesting. I know my histogram in BYEOS shows all the way to the right, almost clipped. Also, the max Normalized value in PI is 0.25. So midway is somewhere around .125. It doesnt need to be exact.

BYEOS stretches the histogram. BYE is effectively the same as looking at the back of camera histogram, which is also rendered from effectively stretched data (processed with in-camera curves). 

 

BYE is therefor lying to you about your signal strength. :p Never trust it! 


  • Dynan likes this

#15 Dynan

Dynan

    Mercury-Atlas

  • *****
  • topic starter
  • Posts: 2704
  • Joined: 11 Mar 2018
  • Loc: NOLA

Posted 19 April 2019 - 11:04 PM

Thanks JR. So can you suggest WHAT I should aim for, which PI scale (Normalized or RGB), BYEOS scale, or any other?

 

Can you straighten me out?



#16 bmhjr

bmhjr

    Apollo

  • ****-
  • Posts: 1185
  • Joined: 02 Oct 2015
  • Loc: Colorado

Posted 19 April 2019 - 11:07 PM

Now I'm confused. I DID expose to the mean of .125 and I got the results you saw in the Dropbox. If I do the same thing, I'll get the same low results. Something isn't meshing here.

 

Should I use the Normalized Real, or RGB balance for the mean?

 

Am I asking the right questions?

Dont worry about the RGB or BYE Histogram, like Jon said it lies.  PI stats for a CR2 file will show 0.25 MAX.  That would be a totally saturated image.  So you want about half way of that at 0.125 for Normalized Scale of a RAW file.  You can use the other stat displays but they wont show as you think.  So 14bit scale would have a max value of 4096, and half of that would be 2048.  If you want to use the 16bit scale the max value is 16384, so a flat looking at that scale would be 8192.  It doesnt matter what scale you use in the stats, I just find normalized easy for me to remember I need a flat that measures 0.125 mean. 

 

There is a pretty wide range where flats should work.  Youre not there yet.  Try doubling your exposure and measure in PI.


Edited by bmhjr, 19 April 2019 - 11:14 PM.

  • Dynan likes this

#17 Dynan

Dynan

    Mercury-Atlas

  • *****
  • topic starter
  • Posts: 2704
  • Joined: 11 Mar 2018
  • Loc: NOLA

Posted 19 April 2019 - 11:25 PM

PI stats for a CR2 file will show 0.25 MAX.  That would be a totally saturated image.  So you want about half way of that at 0.125 for Normalized Scale of a RAW file.  You can use the other stat displays but they wont show as you think.  So 14bit scale would have a max value of 4096, and half of that would be 2048.  If you want to use the 16bit scale the max value is 16384, so a flat looking at that scale would be 8192.  It doesnt matter what scale you use in the stats, I just find normalized easy for me to remember I need a flat that measures 0.125 mean.

That is what I DID (both 14 and 16 bit, and Normalized Real). And you saw the 40% yield of my flats above.

 

Thanks Fellows. I will shoot both flats that yield Half Scale RGB (14 and 16 bit) AND Normalized Real (.125 and ≈.250). Hopefully I'll have some usable flats.

 

I will figure this out...although some confusion remains, experience is a great teacher.



#18 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 23248
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 19 April 2019 - 11:57 PM

Thanks JR. So can you suggest WHAT I should aim for, which PI scale (Normalized or RGB), BYEOS scale, or any other?

 

Can you straighten me out?

On a 16-bit scale, look for your flats to be ~8k DN. Anything between ~6k-~12k should be fine. The key is that you don't clip anything, keep the brightest non-stuck pixels away from the right edge of the PI histogram. 


  • Dynan likes this

#19 TelescopeGreg

TelescopeGreg

    Viking 1

  • -----
  • Posts: 797
  • Joined: 16 Jul 2018
  • Loc: Auburn, California, USA

Posted 20 April 2019 - 12:05 AM

confused...  14 bit image should have a max value of 16k, no?  (Or have astronomers not taken Computer Science 101 yet?)

 

Greg

 

p.s.  My first glance at the subject line read:  Morphine Flats.  I pictured some seedy house on the bad side of town, where a PI (private eye) goes to clear the airy disk in his mind.  There's a great novel buried in there somewhere...


  • Dynan likes 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