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

Photometric color calibration and green cast?

  • Please log in to reply
32 replies to this topic

#1 janame

janame

    Vostok 1

  • -----
  • topic starter
  • Posts: 130
  • Joined: 29 Jan 2019

Posted 27 March 2021 - 12:03 PM

I'm using photometric color calibration on a widefield image of Orion and I'm noticing there's a lot of green left in the background.

In particular, I see what looks like structure between Barnard's Loop and Betelgeuse.  Looking at this [1] image on Astrobin, it looks like there is some OIII in the background, but I don't see anything like the blob in my image, so I'm assuming this is a color calibration issue rather than a real signal.

 

I understand a green cast is common with OSC data, but shouldn't PCC be taking care of this?

 

I adjusted the saturation threshold in PCC, but not the background neutralization because I couldn't find much of an area without stars.  Can I still set a useful ROI on this image?

 

Aside:  I've read that OSC data has a green cast because there are twice as many green pixels.  I don't understand why this should impart a green cast?  It seems like the only effect should be higher resolution in the green channel.

 

Data from ASI2600MC and a Canon 50mm F/1.8 stopped down to F/5.6, stacked in DSS.  In PI: DBE (2 rounds), PCC, Arcsinhstretch (2 rounds)

 

Is PCC+SCNR the right thing to do, or did I make a mistake before PCC that I should address instead?

Would appreciate your thoughts!

 

Green cast:

Green.JPG

Detail of green "blob":

green detail.JPG

With SCNR:

scnr.JPG

 

[1] https://www.astrobin...86367/G/?nc=all


  • klaussius likes this

#2 dswtan

dswtan

    Viking 1

  • *****
  • Posts: 621
  • Joined: 29 Oct 2006
  • Loc: Morgan Hill, CA

Posted 27 March 2021 - 12:38 PM

I've wondered about this too. Airglow, maybe?

https://en.wikipedia.org/wiki/Airglow


  • janame likes this

#3 rgsalinger

rgsalinger

    Cosmos

  • *****
  • Moderators
  • Posts: 8,542
  • Joined: 19 Feb 2007
  • Loc: Carlsbad Ca

Posted 27 March 2021 - 12:50 PM

I zoom in early in the process so I can see the pixels and then use SCNR to kill any visible green cast.  I got that tip from the Adam Block videos. You can wait until later but he recommends attending to at least some of the green cast while the image is still linear. 

 

Rgrds-Ross


  • janame and klaussius like this

#4 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 24,573
  • Joined: 27 Oct 2014

Posted 27 March 2021 - 01:04 PM

PCC doesn't eliminate the green cast, particularly in the background.  You do SCNR first, and then PCC.

 

This is typical PI.  It's like playing chess.  You don't win with one brilliant move, you build your position up slowly, with a series of moves interacting.

 

Fundamental principle.  PI isn't the best because it magically processes better than other programs.  What  makes it unique is the vast array of highly adjustable astro specific tools it puts at your disposal so that YOU can process better.  If you learn how to use them.  I have hundreds of hours in it.

 

The money cost of PI is basically trivial.  The real cost is your time.  In my best year I did 17 images.  The complexity of PI isn't a bug.  It's the main feature.  <smile>

 

Minor point.  Astro OSC cameras use the same Bayer matrix filter as do consumer cameras.  It's not designed for color accuracy, it's designed to be cheap, and to sell cameras.  People buying those cameras tend to like bright greens.  It's a wonder it does as well as it does for astro.


Edited by bobzeq25, 27 March 2021 - 01:11 PM.

  • Tim, acommonsoul and janame like this

#5 janame

janame

    Vostok 1

  • -----
  • topic starter
  • Posts: 130
  • Joined: 29 Jan 2019

Posted 27 March 2021 - 01:12 PM

PCC doesn't eliminate the green cast, particularly in the background.  You do SCNR first, and then PCC.

 

This is typical PI.  It's like playing chess.  You don't win with one brilliant move, you build your position up slowly, with a series of moves interacting.

Why wouldn't PCC eliminate the green cast, though? To me that implies there is an actual green signal coming from somewhere?

 

How do I figure out how much SCNR to apply?

 

Part of the trouble is I have an absurd blue channel in the linear data due to the default white balance in the ZWO native driver. So a linked stretch before PCC is nothing but blue (unlinked is fine).



#6 jdupton

jdupton

    Gemini

  • *****
  • Posts: 3,366
  • Joined: 21 Nov 2010
  • Loc: Central Texas, USA

Posted 27 March 2021 - 01:18 PM

janame,

 

   I would approach this a little differently. I would do a Background_Neutralization first, followed by Photometric_Color_Correction, and then finishing with SCNR.

 

   If you are having gross imbalance of channels from the native driver, fix that first. It is not only messing up the color balance, it is modifying the "supposedly Raw data downloaded from the sensor. That modification of sensor data may be as much to blame as a failure of PCC. You can fix it in the capture software by setting all channels the same. (I think in SharpCap, the setting are in the camera control section and should be set to 50/50.)

 

 

John


Edited by jdupton, 27 March 2021 - 01:18 PM.

  • janame and sbharrat like this

#7 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 24,573
  • Joined: 27 Oct 2014

Posted 27 March 2021 - 01:23 PM

Why wouldn't PCC eliminate the green cast, though? To me that implies there is an actual green signal coming from somewhere?

 

How do I figure out how much SCNR to apply?

 

Part of the trouble is I have an absurd blue channel in the linear data due to the default white balance in the ZWO native driver. So a linked stretch before PCC is nothing but blue (unlinked is fine).

The default settings in SCNR are generally fine.  There are adjustments.  Learning how to use them is a skill to be acquired.  There is no cookbook.  I generally have no need to bother with them.

 

Color in a stack viewed by linked STF is perfectly normal.  There are a variety of causes.   I don't care much about those, though I do mostly attribute it to the terrestrial Bayer matrix filter.

 

You don't use white balance in astro (WB, like the Bayer matrix filter, is a deal for terrestrial folks, terrestrial experience is _often_ not helpful for astro).  You use (after removing excess green with SCNR) PCC to set the white balance.  If you want to pretend you're an alien who grew up under a Type A star, you can do so.  <smile>

 

Other PI tools to tweak color to your taste. Ignore WB before processing.

 

For me the key tools for getting good color with a OSC camera are PCC and arcsinh stretch.  And some secondary tools.

 

Others may have a different strategies.  It's truthfully said, if there are two ways to do something, PI will have five.  <smile>  So different imagers will make different choices.

 

Learning how to process is tough enough.  I don't waste time worrying about causes, if it's not necessary information for processing technique.  I worry about which of the five ways to do something to use.


Edited by bobzeq25, 27 March 2021 - 02:03 PM.


#8 janame

janame

    Vostok 1

  • -----
  • topic starter
  • Posts: 130
  • Joined: 29 Jan 2019

Posted 27 March 2021 - 02:58 PM

janame,

 

   I would approach this a little differently. I would do a Background_Neutralization first, followed by Photometric_Color_Correction, and then finishing with SCNR.

 

   If you are having gross imbalance of channels from the native driver, fix that first. It is not only messing up the color balance, it is modifying the "supposedly Raw data downloaded from the sensor. That modification of sensor data may be as much to blame as a failure of PCC. You can fix it in the capture software by setting all channels the same. (I think in SharpCap, the setting are in the camera control section and should be set to 50/50.)

 

 

John

Unfortunately I was most of the way through collecting these 10 hours of data before I understood the issue and at this point Orion is a next season target. I will definitely have this fixed for the next project, though!

 

Do you find running background neutralization before PCC works better than using the background neutralization as part of the PCC process? I assume you don't do BN and then leave the BN option in PCC enabled?

 

SCNR while still linear?



#9 imtl

imtl

    Gemini

  • *****
  • Posts: 3,491
  • Joined: 07 Jun 2016
  • Loc: Down in a hole

Posted 27 March 2021 - 03:03 PM

Unfortunately I was most of the way through collecting these 10 hours of data before I understood the issue and at this point Orion is a next season target. I will definitely have this fixed for the next project, though!

 

Do you find running background neutralization before PCC works better than using the background neutralization as part of the PCC process? I assume you don't do BN and then leave the BN option in PCC enabled?

 

SCNR while still linear?

I neutralize background during PCC and it works just fine for me. You need to place a proper preview.

 

SCNR could be done at any stage. I honestly rarely use it since I try to balance my RGB to begin with during acquisition. It will depend on sky glow and moon glow. And even your filters+camera transmission. Anyways, more integration smooths things out better than any noise reduction technique.


  • janame likes this

#10 jdupton

jdupton

    Gemini

  • *****
  • Posts: 3,366
  • Joined: 21 Nov 2010
  • Loc: Central Texas, USA

Posted 27 March 2021 - 03:10 PM

janame,

 

Unfortunately I was most of the way through collecting these 10 hours of data before I understood the issue and at this point Orion is a next season target. I will definitely have this fixed for the next project, though!

 

Do you find running background neutralization before PCC works better than using the background neutralization as part of the PCC process? I assume you don't do BN and then leave the BN option in PCC enabled?

 

SCNR while still linear?

   Regarding BN, yes I still leave that option active when doing PCC (or regular Color_Calibration). I treat it much the same way as doing two passes of DBE, one for gross gradients and then again with tighter tolerances for finer non-uniformity left from the first pass. The initial BN gets you close to a neutral background and then as PCC adjusts color, it will also refine the background even more. (At least that is my rationale.)

 

   As Eyal say, you can use SCNR at different places in your workflow, but I usually use it after color correction in the linear state (if I use it all).

 

 

John


  • imtl likes this

#11 janame

janame

    Vostok 1

  • -----
  • topic starter
  • Posts: 130
  • Joined: 29 Jan 2019

Posted 27 March 2021 - 03:27 PM

I neutralize background during PCC and it works just fine for me. You need to place a proper preview.

 

SCNR could be done at any stage. I honestly rarely use it since I try to balance my RGB to begin with during acquisition. It will depend on sky glow and moon glow. And even your filters+camera transmission. Anyways, more integration smooths things out better than any noise reduction technique.

In this case, there are stars and nebulosity just about everywhere. Where would you recommend placing the preview? Is the most empty patch I can find good enough? Looks like in the upper right there are some dark nebulae?


Edited by janame, 27 March 2021 - 03:27 PM.


#12 imtl

imtl

    Gemini

  • *****
  • Posts: 3,491
  • Joined: 07 Jun 2016
  • Loc: Down in a hole

Posted 27 March 2021 - 03:32 PM

In this case, there are stars and nebulosity just about everywhere. Where would you recommend placing the preview? Is the most empty patch I can find good enough? Looks like in the upper right there are some dark nebulae?

In this specific case it might be quite the challenge. The background preview does not have to be a big one. 

The other way with this example specifically, might be to choose a small preview window and balance it by hand as much as possible. I cannot give you some generalized script to run since with "extreme" cases, the way to go is to make up a way to go about it. That's where PI rules in my opinion. 



#13 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 24,573
  • Joined: 27 Oct 2014

Posted 27 March 2021 - 07:39 PM

In this case, there are stars and nebulosity just about everywhere. Where would you recommend placing the preview? Is the most empty patch I can find good enough? Looks like in the upper right there are some dark nebulae?

Don't use a dark nebula.  Zoom in a lot, find a piece of background sky.  You don't need much, 25 pixels X 25 pixels will work.  Maybe 10.  More pixels than 625 is, perhaps surprisingly, not important.  I believe PI also ignores stars for this, could be wrong.


Edited by bobzeq25, 27 March 2021 - 07:41 PM.


#14 jdupton

jdupton

    Gemini

  • *****
  • Posts: 3,366
  • Joined: 21 Nov 2010
  • Loc: Central Texas, USA

Posted 27 March 2021 - 08:07 PM

janame,

 

   I agree with both Eyal and Bob on using a preview for both BN and PCC. If most of the frame is covered with nebulosity, try to find multiple small patches of real background sky (and not dark foreground nebulosity as Bob cautioned). Create the multiple small previews where you find background sky and then run the Preview_Aggregator script (Script | Utilities | PreviewAggregator) to combine all of them. Then use that aggregated preview image as the background reference for both BN and PCC.

 

 

John


  • imtl and jonnybravo0311 like this

#15 tkottary

tkottary

    Viking 1

  • *****
  • Posts: 645
  • Joined: 06 Dec 2015
  • Loc: SunnyVale ,CA

Posted 27 March 2021 - 08:21 PM

Fwiw, I had good success with nuking greens just with pcc using magnitude 15 and aperture 30 under photometry settings .

#16 Madratter

Madratter

    Voyager 1

  • *****
  • Posts: 13,158
  • Joined: 14 Jan 2013

Posted 28 March 2021 - 04:03 AM

If you insist upon using PCC without using the background neutralization, you are going to get color casts in the background from time to time.

 

If you use SCNR to nuke a green cast, by all that is right, please mask the yellows so you don't make beautiful yellows, ugly oranges.

 

The need for SCNR is often a symptom of earlier processing done wrong. I'm not saying I don't use it. I do. But then again, I do things wrong at times too. :p


  • rgsalinger, jdupton, imtl and 1 other like this

#17 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 24,573
  • Joined: 27 Oct 2014

Posted 28 March 2021 - 12:28 PM

 

 

If you use SCNR to nuke a green cast, by all that is right, please mask the yellows so you don't make beautiful yellows, ugly oranges.

 

 

Ah!  I have that problem.  Can you mask yellow in PI?  Could you explain how in detail?

 

OK, found the thread.  Never mind the explanation, here it is.

 

https://www.cloudyni...ip-color-masks/

 

Tried it.  Seems to be excellent.  Do you have preferred settings for which of the three masks to use, amount, layers removed?


Edited by bobzeq25, 28 March 2021 - 12:46 PM.

  • Madratter, imtl and jonnybravo0311 like this

#18 jonnybravo0311

jonnybravo0311

    Surveyor 1

  • *****
  • Posts: 1,711
  • Joined: 05 Nov 2020
  • Loc: NJ, US

Posted 28 March 2021 - 12:54 PM

One of the great things in this community: people are always sharing tips and tricks. Until now, I've just blasted an image with SCNR because, well, it got rid of that excess green. Now, I'll be using the masking technique to protect the yellows. I'm also wondering if I should do the same to protect the teals. After all, the O3 band really isn't the blue so often depicted, but rather a nice teal. I wonder if I could use some pixel math to combine the yellow and teal protection... time to play with some data, I think :)

 

Additionally, I learned about a neat trick for combining previews to use in background neutralization. Thanks John and Madratter for the great tips and tricks. Also, thanks Bob for searching out and linking the thread for the color mask.


  • Madratter, acommonsoul and bobzeq25 like this

#19 sharkmelley

sharkmelley

    Aurora

  • *****
  • Posts: 4,668
  • Joined: 19 Feb 2013
  • Loc: UK

Posted 28 March 2021 - 02:33 PM

Is PCC+SCNR the right thing to do, or did I make a mistake before PCC that I should address instead?

Would appreciate your thoughts!

PCC is great but SCNR is rarely the right thing to do.

 

For instance, take a look at the effect it has had on your image:

 

EffectOfScnr.jpg

 

Attacking the green channel in this manner has turned the background to purple instead of grey.  I've included the histograms underneath.  Previously the histogram peaks were more or less aligned, making the background grey.  After SCNR the green peak has moved a long way to the left, unbalancing the colours and making the dominant background hue quite purple.

 

Mark


Edited by sharkmelley, 29 March 2021 - 12:57 AM.

  • johnpane likes this

#20 rgsalinger

rgsalinger

    Cosmos

  • *****
  • Moderators
  • Posts: 8,542
  • Joined: 19 Feb 2007
  • Loc: Carlsbad Ca

Posted 29 March 2021 - 12:45 PM

Live and learn.

 

Still, I'm a little baffled about one post here. It was suggested to use PCC with a preview. 

 

"I agree with both Eyal and Bob on using a preview for both BN and PCC."

 

How would PCC work properly if it just had a small preview to work from. I totally get the BN part but the PCC part baffles me.

 

Rgrds-Ross



#21 jdupton

jdupton

    Gemini

  • *****
  • Posts: 3,366
  • Joined: 21 Nov 2010
  • Loc: Central Texas, USA

Posted 29 March 2021 - 02:15 PM

Ross,

 

    ACK! You're right.

 

   I was thinking normal ColorCalibration when I wrote that. PCC doesn't use a separate background reference. Good catch!

 

   I still do usually use BN before CC and then also use the background section reference for CC. For PCC, I still find it useful to start out with a reasonably neutral background. I have not done much testing to validate that method for PCC but do see improvements for CC after BN rather than just CC alone.

 

 

John



#22 janame

janame

    Vostok 1

  • -----
  • topic starter
  • Posts: 130
  • Joined: 29 Jan 2019

Posted 29 March 2021 - 02:28 PM

Live and learn.

 

Still, I'm a little baffled about one post here. It was suggested to use PCC with a preview. 

 

"I agree with both Eyal and Bob on using a preview for both BN and PCC."

 

How would PCC work properly if it just had a small preview to work from. I totally get the BN part but the PCC part baffles me.

 

Rgrds-Ross

PCC can use an region of interest for the background neutralization it performs.  I'm assuming this is what was meant by using a preview for PCC.

 

PCC_BN_ROI.JPG


  • bobzeq25 likes this

#23 sbharrat

sbharrat

    Viking 1

  • -----
  • Posts: 602
  • Joined: 28 Nov 2020
  • Loc: NJ, USA

Posted 29 March 2021 - 03:00 PM

Ross,

 

    ACK! You're right.

 

   I was thinking normal ColorCalibration when I wrote that. PCC doesn't use a separate background reference. Good catch!

 

   I still do usually use BN before CC and then also use the background section reference for CC. For PCC, I still find it useful to start out with a reasonably neutral background. I have not done much testing to validate that method for PCC but do see improvements for CC after BN rather than just CC alone.

 

 

John

There is a "from preview" in the background neutralization section of PCC. But that is just used to determine parameters for a ROI within the given image. So AFAIK, you can't somehow use an aggregated preview in PCC. 



#24 jdupton

jdupton

    Gemini

  • *****
  • Posts: 3,366
  • Joined: 21 Nov 2010
  • Loc: Central Texas, USA

Posted 29 March 2021 - 04:21 PM

Shaun,

 

   I agree. I thought I said that in the post you cited but it must not have been clear.

 

   As I said, my brain was stuck on using the normal version of ColorCalibration (CC) when I wrote about using an aggregated preview. PCC can only use a single background POI in the image being calibrated so you cannot use the aggregated preview for PCC.

 

   You can use the aggregated preview for CC which is what I normally do (after applying a BN). I use PCC when I don't like the results from CC.

 

 

John


Edited by jdupton, 29 March 2021 - 04:23 PM.

  • sbharrat likes this

#25 rgsalinger

rgsalinger

    Cosmos

  • *****
  • Moderators
  • Posts: 8,542
  • Joined: 19 Feb 2007
  • Loc: Carlsbad Ca

Posted 29 March 2021 - 04:56 PM

Just to be clear then, you can use the ROI values in the GUI to specify a a preview without stars that PCC can use to figure out what color distortion is present in the image. Then, again to be clear, it will execute a background neutralization for you. So then there's no need to use BN at all if you are using PCC. Have I got all this correct?

Rgrds-Ross




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