Stephan an Deer. Help needed.

[DaveS]

I've been going slowly barking with this.

Captured 15 subs RGB using G2v calibration plus 36 x 10 min subs Luminance with the ODK rig in my sig. Stacked best 90% in DSS with Dark, Flat, Dark Flat and Bias frames, Kappa Sigma.

Resulting stacks pulled into AstroArt 8, aligned and Trichomy before stretching. Applied Histogram Stretch, DDP, and Unsharp Mask to the Luminance.  Applied DDP and Histogram Stretch in that order to the RGB before LRGB synthesis (Lab colour protocol). Cropped off the edges before Gradient Reduction, Adaptive Subtract. Pushed the Histogram and colour around a bit more. Finally saved as JPEG. The PNG is 70 MB. 

Need to get rid of an overall blue-green tinge to the background and a greenish tinge around the bottom of NGC 7331, which also has a burned out core, even more obvious in JPEG. There are also two or three light streaks coming i to the upper right, possibly from stars just outside the field.

Any ideas? I could post up the unprocessed stacks but they are a bit big.

 

[bottletopburly]

Upload to Dropbox then share link .

[DaveS]

Let's see

Stephan and Deer

[DaveS]

Hmm...57 views an only one other reply. Ah well.

In the absence of any other suggestions I've been kicking the data around a bit more. Slowly getting there I think.

The latest iteration. The core of NGC 7331 is less burned out, but I'm still not convinced about the colours.

 

[Laurin Dave]

Hi Dave, I ran it though Pixinsight as follows: -  register, crop, RGB channel combine, Dynamic Background Extraction on both , Photometric colour calibration, ArcSinh stretch on RGB,  Histogram Stretch on RGB then colour boost in Photoshop using LabColour, HS on Lum and then LRGB combo...   came out looking like this...

Galaxy and star colours look ok to me.  I'd suspect that its your colour calibration that's gone astray although as I'm not familiar with AstroArt I can't suggest how to fix it.   The light streaks are puzzling as the nearest star in the direction they are coming from is HD214667 which is mag 9, Eta Pegasus could be the culprit at mag 3 but it is a long way off.  I'd check that they are on all your subs and if not would suspect something closer to home.   Focus quality also looks a bit soft and different between filters with green best and blue,  I'd measure all your subs to see what's going on with focus.

HTH

Dave

[DaveS]

Thanks Dave.

That light streak is on all my subs but worse on the Luminance. I will recheck my focus, but the last time I tried the difference between filters ewas less than the difference between focus runs. Perhaps I will check with a Bahtinov mask on the next clear night (This year? Next year?? Sometime??? Never???!!!).

I calibrate my monitor using a ColorMunki, and it's a wide gamut job.

I had a similar streak on my images of NGC 3718, but only a single streak rather than the 2 or 3 on this image. Thought it was due to a star outside the field then, but not sure now. May need to buy a dew shield for the ODK to give a bit more screening. Not cheap though.

[tomato]

Hi Dave, only just seen this thread, I'm fed up trying to process my M31 data so I gave yours ago through my usual workflow, thanks for posting it.

Analysed, registered and normalised in APP.

Channels loaded into Startools v1.8 and progressed through the default workflow. Then adjustments made in AP. I cant get rid of the aretefacts in the top left, but to my inexpert eye, there is good colour and detail in NGC7331 and the Stephan's Quintet. I have similar flares from bright stars just out of the FOV, but never multiple ones as visible here.

[DaveS]

I'll have another go at this tomorrow (Maybe, I have other things to do as well), taking more care over the colour balance. Not much I can do about the streaks, maybe a masked inverted stretch.

[DaveS]

Thrashing around with this image. This is where I've got so far. The file name says "Fourth LRGB" but it's the second or third iteration I've done today. The others weren't worth saving. Still having trouble retaining detail in the core of NGC 7331 and getting good colour. All my efforts have a bright burned-out dot in the centre, suspect saturating the CCD.

[DaveS]

I dunno but I think this might be the best of my iterations. This base file name is "Fifth LRGB". There are variations of "Sixth (Or should that be "Sith") LRGB, but think this might just be better. This is from a stack in AstroArt 8, the sixth is from a stack in DSS. But I'm about ready to call a halt to going round in circles. Maybe if I manage to collect more data I'll be able to progress. I don't have any H-alpha but don't know how much effect it will have at this scale.

 

[tomato]

I think your final version is much better than your original post so your time and effort has been worthwhile. 
I know the feeling of going around in circles, I guess one measure of processing skill is getting close to a result you are satisfied  with but needing fewer and fewer iterations.

I still need all of the UK cloudy nights to get somewhere close.

[DaveS]

18 hours ago, tomato said:

I think your final version is much better than your original post so your time and effort has been worthwhile. 
I know the feeling of going around in circles, I guess one measure of processing skill is getting close to a result you are satisfied  with but needing fewer and fewer iterations.

I still need all of the UK cloudy nights to get somewhere close.

Thanks, yes I think this is (Just about) the best so far. Really need to hone my processing skills then I won't be going round in circles so much, guess I'm still learning. This image had rather a lot of magenta in the background and spiral arms, but the latest AstroArt has an "Attenuate single colour" function in the colours menu that I've been experimenting with. Seems to be quite useful.

[wimvb]

I just downloaded your masters, and it seems a very nice dataset. In the luminance data, there is inverted vignetting, so you might want to check your flat calibration. With cmos cameras inverted vignetting can be caused by inconsistent darks or bias being used in stead of flat darks. I don't know what might cause it with CCD cameras. Too bright flats perhaps?

The core of ngc 7331 definitely is not burnt out, but it requires carefull stretching. In your original image, you may have brought the black point and white point in too far.

I'm working on the image now, and will report back when I've processed it.

[wimvb]

Here's my result.

I think you managed to capture some IFN, but I found the background a bit too uneven in RGB to enhance it.

The stars in your image are quite large, and oversampled. With a 2 m focal length and 6 micron pixels, your image scale is about 0.6 arcsecs/pixel, which I think the sky can't really support. If you bin 2x2, even for luminance, you will probably get better subs without losing detail.

Edited October 21, 2021 by wimvb
fixed blocky stars

[DaveS]

Thanks Wim, that's a much better result than I got. I think I'm inclined to agree with you regarding the sampling, my sky can't support 0.61"/px unfortunately. I may try software binning and see how it goes.

I think this data set was from stacking in DSS using Bias as well as the usual Darks, Flats, and Dark Flats. This might have upset the calibration (DSS complains if you don't include Bias). I have a later set from stacking in AstroArt 8 which I'll upload tomorrow.

Following @Laurin Dave's query about the focus I ran a check with a Bahtinov Mask but could see no shift in the diffraction figure between L, R, G, or B. Maybe higher order aberrations are in play here. I may have to check the mirror spacing but that's a real pain to do.

[wimvb]

6 hours ago, DaveS said:

I may try software binning and see how it goes.

With your ccd camera, binning during data collection will give the most benefit in terms of noise reduction of course. If you use an auto focus routine, the hfr or hfd measurements normally give you an indication of sky quality. (Hfr is half flux radius, hfd is half flux diameter. Hfd is similar to fwhm, full width at half maximum.) I find that these values can differ quite a lot. On nights with high hfr values, you'd use bin 2x2, while if you experience exceptionally crisp and clear nights with good seeing, you could do unbinned data collection. Although even then binned 2x2 may give you enough details.

The masters you provided where very clean as regards to noise, and I didn't have to do any noise reduction, other than my usual chroma noise reduction of the rgb image before lrgb combination.

[DaveS]

Ah, was replying just before going to bed. What I was meaning was software binning the data I already have. I will hardware bin any new data.

[wimvb]

I understood as much. It will be interesting to see the stacked binned data, but I think that quality wise the improvement will not be that obvious. I found your master images to have a good signal to noise ratio already.

[DaveS]

Looking again at your process the core region of NGC 7331 shows much better detail than mine, plus the warmer colours are more pronounced, minse are a bit washed out and tending to blue. I'm thinking you used PI? I am still seeing an annoying dot in the centre which leads me to think it's saturated somewhere.

[wimvb]

1 hour ago, DaveS said:

I'm thinking you used PI?

Yes

1 hour ago, DaveS said:

I am still seeing an annoying dot in the centre which leads me to think it's saturated somewhere

It's possible that I stretched the core to slightly past its naximum, but the linear image definitely wasn't blown out. What I wanted to avoid is this:

https://pixinsight.com/examples/NGC7331-CAHA/index.html

And as long as the core isn't blown out you can always combine a lesser stretched copy with a more heavily stretched one.

[DaveS]

Sorry, I was referring to my images, where I have a solid dot in the centre that just doesn't want to play nice. I really must take more care over my phrasing, the core in your version looks better than all of mine, though the latest version of AstroArt is very capable. I really don't want to get involved with PI, the last time I tried  it I nearly had my brains trickling out of my ears.

Going to pause again and have a think about how to proceed.

[wimvb]

The workflow I used was roughly this:

• Aligned the masters
• Cropped all the masters to remove stacking edges and the worst of the light intrusion
• Combined R G B

RGB:

• Flattened the background with Dynamic Background Extraction
• Neutralised the background 
• Colour calibration using photometric process 
• Colour noise reduction in the background
• Arcsinh stretch which preserves and boosts colour
• Increased colour saturation in ngc 7331

L:

• DBE to flatten the background
• Deconvolution to  increase sharpness a tiny amount and shrink the stars ever so slightly
• Brought in the black level to just before clipping
• Histogram (levels) stretch with the midpoint slider set to 0.25. Applied this several times until the core was getting very bright. Doing the stretch in small steps gives better control
• HDR compression (HDRMT) to manage the bright core of ngc 7331
• Continued histogram and curves stretch
• Local contrast boost using Multiscale Median Transformation to enhance the spiral arms in ngc7331 and give a little more detail in the quintet
• Bring the white level down to 0.85, meaning that everything that was pure white (star cores, intensity level 1 in PI) now is at 0.85.

LRGB combination

• Blurred RGB and applied a 20% mask
• LRGB combination with colour saturation increase
• Blurred the image but less than before with the mask disabled
• LRGB combination with the 20% mask.

Repeated this several times. This controllably adds luminance to the colour data without washing out the colour.

Finished off qith traditional star reduction (morphological transformation)

Although this workflow is specific to PI, it can be adapted to other software.

Hope this helps.

Edited October 23, 2021 by wimvb

[DaveS]

Many thanks Wim.

Yes, I can follow this though AA has some differences. I may try stretching in Affinity as it has a mid-point slider which AA doesn't.

I've been trying various different routes through the processing, all the way from the original subs. Trying both DSS and AA for stacking (With and without Bias, think without is better, maybe). Then a simple Linear Gradient Reduction (Which doesn't mind alignment edges) for the masters. Then either Trichromy before any stretching or stretching before Trichromy. At the moment I'm finding the Digital Development Process in AA useful as it applies a log stretch. During this I keep an eye on the histogram to avoid Black and White Clipping. For the Luminance I've also been applying a touch of Unsharp Mask though this has a tendency to white clip. And so on, and so forth. Because AA doesn't like combining different sized frames I wait until after LRGB combination before cropping and Adaptive Gradient Reduction, then colour curves and maybe another Histogram Stretch.

AA doesn't have an Arcsinh stretch unfortunately, else I'd be using it and saving myself a lot of trouble.

Beginning to think I'm going all round the houses to get to where I should be able to more straightforwardly.

Time to look again at PI? Possibly, though I ran out my trial version and had to uninstall it when my old boot drive became full. Have a new boot drive so have room now. I also found it unintuitive compared with AA.

[wimvb]

22 minutes ago, DaveS said:

I also found it unintuitive compared with AA

Different image processing packages have different philosophies. In PI it helps to think in terms of levels of brightness, levels of detail (1 px level, 2 px level, 4 px level, etc) and to separate luminance from colour (chrominance). And it helps if you have a decent knowledge of math terms, because the processes rely on strict mathematics, and the process names refer to the math that is used. I guess that's why people tend to be put off by it. PS on the other hand comes from image manipulation with strokes, air brush techniques, and applying filters (among many others), so you refer to areas of an image that need adjusting. In the end it's very much down to understanding the tools, how they work, and getting better at applying them.