Green image from ASI 071 pro - Help needed!

[gorann]

Wednesday night I was imaging the Cocoon nebula with my Esprit 150 and new ASI071 pro. The object was near zenith, no visible clouds or moon, and a nice Milky Way accross the sky. My SQM showed 21.1. Even guiding was good (0.5"RMS). So everything seemed fine until next morning when I started processing the subs. The initial processing (calibration, debayer, aligning and integration) was done in PI and it was when I did the debayering that I got the chock. Everything was extremely green. I used the matrix setting given by ASI for this camera which is RGGB and I have used it before and never seen this green before. If anything I have had a blue cast, but I have only used the camera four nights. So I tried all 8 different matrix settings in PI and got various terrible colour casts and none better than with the RGGB - and it seems odd that the RGGB matrix suddenly should have changed.

Wim @wimvb suggested to me that it could be Northern light that I could not see by naked eye. However, the green cast was identical in all 81 subs stretching over 6 hours, and I would expect that Northern light varies in intensity.

After much tweaking with colour curves and other colour adjustments in PS I finally reached a presentable image but it was a heck of a job and I initially gave up for a while. Here is one of the green subs (only given a stretch) and also what I finally rescued after much processing.

So please help! Could I have set something wrong in ASICAP. The only thing I had intentionally changed from previous sessions was to increase offset from 20 to 30. Gain was 200 and exposure 5 min but I used that before. Saturday night is supposed to be clear and now I dread what I will find on Sunday morning.

[vlaiv]

I think offset might have something to do with it. But let's put that aside for a bit and see if anything else could be causing this.

Did you change filters you use? Did you use LPS filter earlier but decided not to use it now? Did you apply UV/IR cut filter? ASI071 has AR coated window, so you should probably use UV/IR cut filter. Maybe there was something with filters that caused this.

If everything is exactly the same as before hardware wise, next thing to look is at software. Did you change any software components (like capture program, or driver settings other than offset? Did you manipulate color balance in any way?)?

Now onto offset. If you look at ASI071 specs, please note QE graph - I think it is the key to understand what is going on, so if your sky is very clear with low LP, it should have slight blue tint to it naturally, but camera might not see it as blue because of needed color balance. Setting offset too low can alter dark areas (where there is no signal, or very weak signal from natural sky glow). It can cause different color artifacts depending on how much there is clipping to the right.

This will not usually happen if you have more LP, since LP signal will accumulate enough to overcome clipping (it will push histogram to the right), but in very low LP and in very transparent skies there might not be enough signal, and clipping can occur.

Can you take one of green frames (a good one that you would use as a reference frame), and one of your previous frames that did not have this issue. Both should be uncalibrated frames, straight out of camera (if you happen to have one of previous, and keep raw / uncalibrated versions of files), and do detailed histogram on both of files so we can compare.

Or you can post one of each here, so we can have a look and try to diagnose what might happen.

In the mean time, I don't think that 20 or 30 offset is high enough, but you can check this. Take a set of bias files, and see if you have any zeros (zero values) in any of them - You can do this by taking for example set of 32 bias frames and do simple stack by min value, and look at minimum pixel value for such stacked image - it should not be 0, it should be at least >4 for proper offset.

To explain how offset can impact background color cast, look at this:

With old offset, very low values were clipped to 0, so background had roughly uniform R, G and B components

Now you changed offset to new value, red and blue are still clipped over whole range, but green because it has higher QE over whole spectrum is no longer clipped and remains non zero, so R:0, B:0, G>0 = green cast.

If you lower offset to proper value (so there is no clipping), you will still have a bit of color cast, because both Red and Green do have slightly higher QE than blue - this should be properly handled by color balance though (as no clipping and nonlinear data change will happen).

[gorann]

Thanks a lot Vlaiv! Nothing in my hardware was changed and like before I used a Baader UV/VIS filter (the one used for Lum for mono imaging). The only software change I did was to increase offset from 20 to 30 and I hope you are right in suggesting that this may be the casue of my problem.

I attach four RAW (16 bit) files as they come out of the camera (with the txt files for info). Those are the relatively comparable ones I have. It would be wonderful if you could figure something out from these!

1) From my "green session" with 3 min exposure time. Gain 200, offset 30

ASICAP_2018-09-12_22_12_40_831.FITASICAP_2018-09-12_22_12_40_831.FIT.txt

2) From my "green session" with 5 min exposure time. Gain 200, offset 30

ASICAP_2018-09-12_23_56_08_563.FITASICAP_2018-09-12_23_56_08_563.FIT.txt

3) From 1 September Lum filter with 3 min exposure. Gain 150, offset 20

ASICAP_2018-09-01_23_27_30_896.FITASICAP_2018-09-01_23_27_30_896.FIT.txt

4) From 2 September Baader Ha filter, 5 min exposure. Gain 200, offset 20

ASICAP_2018-09-02_22_50_35_428.FITASICAP_2018-09-02_22_50_35_428.FIT.txt

 

 

[wimvb]

My bet is on a natural cause, either aurora or air glow, similar to aurora.

https://www.universetoday.com/112237/how-to-see-airglow-the-green-sheen-of-night/

The viewing angle seemingly doesn't quite match, but remember that your sqm read 21.1 at the time of occurence, which is very dark to start with. And the article is about seeing it, which requires a much higher intensity. For that, the light must be "denser", which you get at lower angles.

Light pollution can be a bummer, can't it. ?

[gorann]

With the first sub tonight I will set offset to 30 and then go back to 20 for the rest. That should prove or disprove the offset hypothesis, and if it is not green at offset 30 then in was probably Aurora Borealis or something equally nasty, If they are all green again then that aurora must have parked over here or there is something seriously wrong with camera or software.

[vlaiv]

This is very confusing and I can quite figure out what is going on.

Left histogram is for 3minute gain 150, offset 20 no green cast, right one is for 3 minute, gain 200, offset 30.

It is interesting that left histogram, at lower gain setting and smaller offset shows significantly higher background pixel values, and uniform single peak - that is something one would expect from mono camera, not osc, or from OSC that has been very properly white balanced.

Right one shows three distinct peaks, as one would expect for raw / unaltered data from OSC camera (different sensitivity in R, G and B).

What confuses me is that right image shows much lower background pixel values although it has higher gain and greater offset. This simply does not make sense, and it looks like first image has been either shot in much stronger LP, or has been altered somehow (aligned color peaks, greater pixel values).

I see no sign of clipping due to offset, so I don't think change in offset is the reason. I also don't think that Northern light is to blame - then green image (right) would have larger background pixel values consistent with stronger LP.

Ah, found it as I was typing this - very simple solution to this mystery, but it shows that you indeed changed other parameters than offset alone. Here is screen shot from FITS header info for both images:

First image has been white balanced in capture application. This is consistent with above histograms (red and blue multiplied with some value - higher pixel values, matching R, G and B histogram peaks) and explains lack of color cast in first sub.

So nothing wrong with camera, offset or anything - if you shoot totally raw without internal camera WB - you will get green tint simply because camera is the most sensitive in green like above QE graph shows.

[gorann]

13 minutes ago, vlaiv said:

This is very confusing and I can quite figure out what is going on.

Left histogram is for 3minute gain 150, offset 20 no green cast, right one is for 3 minute, gain 200, offset 30.

It is interesting that left histogram, at lower gain setting and smaller offset shows significantly higher background pixel values, and uniform single peak - that is something one would expect from mono camera, not osc, or from OSC that has been very properly white balanced.

Right one shows three distinct peaks, as one would expect for raw / unaltered data from OSC camera (different sensitivity in R, G and B).

What confuses me is that right image shows much lower background pixel values although it has higher gain and greater offset. This simply does not make sense, and it looks like first image has been either shot in much stronger LP, or has been altered somehow (aligned color peaks, greater pixel values).

I see no sign of clipping due to offset, so I don't think change in offset is the reason. I also don't think that Northern light is to blame - then green image (right) would have larger background pixel values consistent with stronger LP.

Ah, found it as I was typing this - very simple solution to this mystery, but it shows that you indeed changed other parameters than offset alone. Here is screen shot from FITS header info for both images:

First image has been white balanced in capture application. This is consistent with above histograms (red and blue multiplied with some value - higher pixel values, matching R, G and B histogram peaks) and explains lack of color cast in first sub.

So nothing wrong with camera, offset or anything - if you shoot totally raw without internal camera WB - you will get green tint simply because camera is the most sensitive in green like above QE graph shows.

Thanks Vlaiv,

but now I am confused since I did not change any of those settings (and do not know how to do  it). Should I  find out how to change them back to the original white balance or should I continue shooting green images. I am setting up for the night right now!

[vlaiv]

Are you using ASICAP software? It looks like settings should be on right panel, probably not visible immediately, but accessible thru "..." button (same place you changed offset?):

It looks like there are two modes for WB - auto WB and manual WB settings. From image above, you can see what WB settings you used in first image. So WBR 60 and WBB 99, but we don't know if it was manual WB or auto WB.

As for which one you should use, well that is up to you really, and the way you process your images. I tend to turn off WB settings (no auto, and everything on 0) and correct color in post processing. This is because I've got a lot of LP and there is always some sort of color cast - which I like to handle myself rather than leave it to software on auto mode. I also don't like any additional changing of data in camera since it is 16bit format, and prefer doing any sort of pixel manipulations in 32bit.

[gorann]

Thanks a lot again Vlaiv! You seemed to have solved the problem.  I now have it on the default 60/99 and started shooting on the Ghost nebula.

[gorann]

2 hours ago, vlaiv said:

Are you using ASICAP software? It looks like settings should be on right panel, probably not visible immediately, but accessible thru "..." button (same place you changed offset?):

It looks like there are two modes for WB - auto WB and manual WB settings. From image above, you can see what WB settings you used in first image. So WBR 60 and WBB 99, but we don't know if it was manual WB or auto WB.

As for which one you should use, well that is up to you really, and the way you process your images. I tend to turn off WB settings (no auto, and everything on 0) and correct color in post processing. This is because I've got a lot of LP and there is always some sort of color cast - which I like to handle myself rather than leave it to software on auto mode. I also don't like any additional changing of data in camera since it is 16bit format, and prefer doing any sort of pixel manipulations in 32bit.

I have now checked my first sub and setting the white balance at default 60/99 (with offset at 30) gave me back the pinkish tint I used to have!

Thanks again!

 

[wimvb]

Good thing I was wrong, because air glow / aurora is a lot harder to fix. ?