RGB issues - first light ZWO 1600MM

[mgirdwood]

Hi 

Please can anyone point me as to how i made some terrible errors in last nights first light with1600MM pro.
TS Quad 65mm 
100 seconds at 111 gain ( i think) 2x2 binning 30 x each  RGB attached , haven't added any Lum
image is stacked with no flats ( looks the same with them ) but i was trying to see if the flats was the issue. The Flats do get rid of all the dust bunnies
STF stretch nothing else on first and a quick ABE subtraction on second then STF stretch

thanks for looking as i would really like to get the Mono sorted out  - Mark

 

Edited November 6, 2020 by mgirdwood

[knobby]

What did you stack with ? DSS always looks washed out and very dark till stretched and tweaked ... The gradients are most likely caused by the moon .

[mgirdwood]

Hi, tried stacking both in DSS and Pixinsight
The Moon was below the horizon for much of the run, would the Moon account for the very bad colour integration though  ? 

[wimvb]

The gradient can come from light pollution. You may need to do background extraction in two steps: first using division to get rid of vignetting, then subtraction to get rid of the colour gradient. Use DBE, not ABE, because it’s easier to control.

[ollypenrice]

Flats with a CMOS camera need dedicated flat darks taken at the same gain and binning as the flats themselves.

Could you clarify what the top and bottom images are in your post? I'm not sure what I'm looking at.

Olly

[blinky]

I dont bother with flat darks Olly and dont have any problems with my ASI1600

[david_taurus83]

31 minutes ago, ollypenrice said:

Flats with a CMOS camera need dedicated flat darks taken at the same gain and binning as the flats themselves.

Could you clarify what the top and bottom images are in your post? I'm not sure what I'm looking at.

Olly

The first one is just an STF stretch and the second has had an iteration of ABE.

@mgirdwood
did you apply a full set of calibration frames? Darks for the lights, flat darks for the flats? All at 2x2 as you say you binned?

[blinky]

I think you maybe had some cloud/fog and thats whats caused the issue

[ollypenrice]

We need to be clear what the issues are. I don't see anything odd about the re-green gradient, red left, green right. We get those here even with virtually no LP. I often got them either side of a diagonal with my OSC CCD as well. DBE should easily fix that. (ABE won't do a good job on galaxies. It tends to put markers too close the galaxy and introduce dark/light rings around it as it compensates.)

But why haven't the flats taken out the vignetting? Do they usually work properly?

Olly

[wimvb]

19 hours ago, mgirdwood said:

image is stacked with no flats ( looks the same with them ) but i was trying to see if the flats was the issue. The Flats do get rid of all the dust bunnies

 

1 hour ago, ollypenrice said:

But why haven't the flats taken out the vignetting? Do they usually work properly?

Olly, the way I interprete the original post is that no flats were applied here. When flats were applied, they removed dust shadows.

1 hour ago, ollypenrice said:

We need to be clear what the issues are.

That's what is confusing me (too). I agree with your comment that ABE should not be used here.

[mgirdwood]

Thanks Everyone for the replies. Apologies for poor description - i was a tad frustrated after a fair few hours trying different methods.

Coming at this from a OSC the processing i realise will be quite different at least to begin with.

With a clearer mind i re ran the PI older Batchpreprocess with all RGB flats and set of darks -  all at 2x2binnig

I think where i am most unsure is why i am getting the large colour gradients across the image, from the first suggestion i reduced the number of subs to those taken before moonrise to see how much difference it might make.  these were results today.  first is STF stretch of the integrated image and the second is with DBE (Division and subtraction).  19 RGB x 100 sec at -15.   My skies are approx Bortle 4 i believe.      

 

[AKB]

Taking your first image in the previous post above, applying DBE, and a bit of colour balance and stretching, I get this.  It seems not too bad to me?

Edited November 7, 2020 by AKB

[ollypenrice]

That looks as if it should be fine. Your DBE-adjusted image shows that the gradient is well within the range DBE that can handle. You just need to refine the way you're using it.  As it is you've got a pretty even background, as AKB's application shows.

Since you're only attacking a colour gradient, now, I would suggest subtraction of the gradient model, not division. Sky gradients are additive, as Wim said earlier. Personally I would not place many markers  (maybe 12?) and I'd keep them well away from the galaxy. (That's essential.)

I find I rarely need to adjust the background sky colour after DBE but, if you do, you could try both SCNR green and Background Sky Neutralization to tweak it.

The data will work!

Olly

[wimvb]

6 minutes ago, ollypenrice said:

Personally I would not place many markers  (maybe 12?)

Yes, that's how I would do it. For vignetting gradients, one sample in each corner, one further in from each corner along the diagonals, and one along each edge. Make the samples larger than the default 5 pixel radius. I sometimes go up to 35. Stay away from large stars, but don't mind covering smaller ones. And as Olly already warned for, don't even get close to the main object. (You can always take an image from the internet as a reference.) Increase tolerance until sample weight (next to the sample image) is larger than 0.75 for all three colours. The sample image should look as neutral light gray to white as possible, without a distinct colour cast.

In the correction section, if you leave "normalization"(?) unchecked, you will get a neutral(ish) background. Personally, I have this option checked, in which case any colour cast is left as it was. I like to keep background neutralisation separate from gradient removal. But that' s just me.

For a linear gradient, you need fewer samples, about 3 along the gradient and 2 - 3 perpendicular to it.

[mgirdwood]

Thanks Everyone, i appreciate the helpful information and will see if i can work on the image with these suggestions.  LRGB is going to be a new experience, i just hope i am up to succeeding even a little as want to head over to narrow band at some point

Mark

[wimvb]

54 minutes ago, mgirdwood said:

Thanks Everyone, i appreciate the helpful information and will see if i can work on the image with these suggestions.  LRGB is going to be a new experience, i just hope i am up to succeeding even a little as want to head over to narrow band at some point

Mark

LRGB isn't that difficult if you adhere to the principle that you process the RGB image to get maximum colour impact (level of saturation) and the L image to get maximum contrast and detail. LRGB combination will then combine these two: detail and contrast with colour impact. In an OSC image, it's more difficult to separate colour from lightness, because they are both in the same image. in LRGB imaging, colour and lightness are separated and you optimise each before you combine them. Once you get used to this, LRGB processing can in fact be easier than OSC processing.

[mgirdwood]

I still see this taking me quite some time to get the basics sorted  - but i do appreciate the encouragement.  (always have the OSC to fall back on    )

Cheers everyone.. 

 

[david_taurus83]

Have you run DBE on the individual channels before combining into a colour image?

[mgirdwood]

No only ran DBE on the combined image  - i can certainly add the individual DBE into the workflow if its a better option