5.76
#...
·
·
2
likes
|
|---|
|
I'm pretty happy with the level of detail my SVX152 managed to find here (with a nod to the AP1100AE for good guiding also). The number of little galaxies in the background amaze me. I welcome any comments on the processing. A lot was aimed at bringing out the tidal tail. It is R, G, B, a synthetic Luminance from the mean of those, with a bit (20%) of Ha added in later to provide some of the globules of star forming regions and a bit more red central cores. HDRMt and LHE bought out much of the fine detail in the galaxies, BlurXTerminator was used primarily on the stars, with a bit of non-stellar. This was drizzle integrated the reduced back to the original 0.62"/px. I cropped to show the three galaxies but also the tidal tale, so the bottom edge is the edge of the frame. I probably should have left a bit more room there but frankly did not expect to get the tail and even if I did, was not aware it was that long. Criticism and suggestions welcomed. Linwood https://www.astrobin.com/full/fpfag2/0/ |
9.82
#...
·
·
1
like
|
|---|
| My only glaring issue is the level of noise in the background. You might benefit from dumping some time into luminance. The idea is luminance is where all your detail comes from. Most folks dump 3 times the luminance of your RGB channels... The reason is you get a clean signal with less noise. Doing a synthetic luminance requires more denoise than normal. I literally just kinda did something very similar and didn't shoot lum and kinda regret not dumping the time into it. Personally I didn't find any gain in a synthetic luminance layer, and I ended up processing without. In your case, what I would do is extract the stars (some galaxies may come with extraction but that's okay) then apply a mask to protect the galaxies in the starless image, and use some heavy denoise on the background. Throw it all back together. The tidal tail will remain intact as well. Neat trick I just learned is using MMT to create a mask around the stars/galaxies allowing you to bring out detail in the background after denoise. I'm being very brief about the technique since I'm on the phone, if you want more detail, lemme know. I just don't know your level of experience. |
|
5.76
#...
·
·
1
like
|
|---|
|
Thank you @Brian P . I used NoiseXTerminator, and used it very lightly. I got almost no noise by turning it up, but was afraid it looked too processed. Maybe I under-estimated. But thanks for the comments on Luminance, I used to do Luminance and was swayed by some experimentation that I got better detail with individual filters, perhaps due to better color alignment. Though my experimentation was more on the C11 which is not as well color corrected, I suspect, as the SVX152T. I should probably try again. Thank you. |
|
9.82
#...
·
|
|---|
Linwood Ferguson: NoiseX is pretty powerful. I'd probably start at 0.5 on the background of this image. It doesn't have to be completely noiseless... Honestly the rest of the image is pretty solid. |
|
5.76
#...
·
·
1
like
|
|---|
|
I decided to try a bit heavier NoiseXTerminator. It removes the small scale noise, but not the splotchiness. It also yields what I thought was too smooth of a look, too artificial, but curious what others thing. Getting rid of the splotchiness would take something more invasive, maybe masking off only the background carefully and applying a blur.  |
3.61
#...
·
·
1
like
|
|---|
| Great detail in the galaxies Linwood!! Agree on the noise a bit. What are your settings in NoiseXT? I do a very light pass in linear just before stretching to knock the fuzz off, then a little heavier at the end of nonlinear. Have you tried Bill Blanshan's star reduction scripts? On a very light setting might take the halos down just a bit in some of the brighter stars. Overall you pulled out some great detail in the galaxies and I really like the color treatments. Just some ideas you are way more expert than I. |
3.81
#...
·
|
|---|
|
Have you by any chance used noisex or blurx in the linear stage? I tend to get blotchy backgrounds if i use one of the two tools before stretching. Just a personal observation on a few targets. anyhow, your image looks great! cs andi |
|
5.76
#...
·
·
1
like
|
|---|
Andi: I always use BlurXTerminator right after stacking, or at most after DBE. NoiseXTerminator I use late, well after stretching (it says it works in both linear and stretched and seems to). I struggle a bit with the interaction between BlurXTerminator and HDRMT/LHE, I think you can apply a mild BX, but then both of these subsequently work more strongly than they would otherwise capitalizing on the contrast from BX and making images look a bit over-baked in places, notably galaxy dust lanes. I tried here to be more careful, and need to go back and redo some prior ones. What you see here in the dust lanes is attributable maybe 1/3rd to each, and still at some zooms looks rather over-baked even though I tried to be careful. But noise... still debating if I should just run NX again, take out some of that color noise, but leave the splotchy background. Taking that out, even if I can, I have to go back much further I think. |
|
3.61
#...
·
·
1
like
|
|---|
| Linwood, try running NoiseXT just before stretching. After DBE, color calibration, and BlurXT. Use a light setting, say 0.75 strength and 0.15 detail. |
|
5.76
#...
·
·
1
like
|
|---|
Jeff Horn: Thanks. I shall. Not sure I want to go back that far on this one right now (I tend to get burned out on one subject), but I have a redo of M81/M82 I need to do that is similar in many ways. |
11.99
#...
·
|
|---|
|
So - first of all - it is refreshing to see an image like this where the imager has made the effort to transport a huge scope and heavy mount to a dark site, at a time when so many images seem to be from data acquired from remote observatories. I love the detail in all three galaxies, down to the fine H-alpha regions. I can't remember when I have seen these better resolved. Yes, true luminance would have helped in bringing out the tidal tail even more, perhaps in increasing the SNR to help BlurX extract even more detail, but this is already very good. The noise did not distract me one bit - there are far too many things that are good about this image. |
15.85
#...
·
·
1
like
|
|---|
|
First of all, Great job Linwood! The detail and resolutions is excellent in the galaxies. Second, I agree with pretty much everything here, however I’m going to add few things so please don’t take it the wrong way! Noise: I think there is definitely more noise than I prefer but I think this is somewhat of a personal taste, at least I feel it is. I do not like noise and have a tendency to smooth out things more than this. As others have mentioned this can be done in several ways but in reading your processing tools I would do everything in the liner stage. Both BlurX and NoiseX are designed to work together both before the image becomes stretched. There are several threads that have been posted here on AB that really has a good work flow for the 2 processes. Of course you’ll need to play with the settings to suit your taste. Here is my personal work flow prior to stretching an image. Slight Crop DBE Image Solve/SPCC BlurX NoiseX StarX Once I have both the starless & star layer I then process them accordingly. Blotchiness: I agree and see this and I think @Jeff Horn is on the right path for you which goes back to what I mentioned do your X processes prior to the actual stretching. Color: I like the color you have and I think they are spot on, but I also think your background is “over saturated” here and that could possibly be contributing to the blotchiness or at least making it more prominent. A couple suggestions is to create a mask however you prefer making your masks, (I like the Game Script) and apply it to protect your background and then work your galaxies. As for now if you want to adjust the background (without redoing your processing) do the reverse. Apply the mask to protect the galaxies and then in curves pull down the curve at the mid-background to background level. As a side note, if you look closely at the image there are 3 main areas that has a heavy magenta color to it. (In looking again, it’s in the whole pix) Across the bottom as well as what I’ll call a hot spot just to the upper right of center. With the mask applied you could also use “correct magenta stars” script and that can help as well but play with it and I think you will see what I’m talking about. Once you do that you can remove the mask and adjust according to your tastes. Again please do not take anything I have said the wrong way! This is an excellent image with a wealth of detail! These are just my suggestions and also my personal workflow prior to stretching any of my images. Dale |
0.00
#...
·
|
|---|
| Make sure when using NoiseX in linear that you use the 24-bit STF lookup tables. |
|
5.76
#...
·
|
|---|
| @Arun H I need to try that, thanks for the pointer, never used ACDNR that I recall. And yes, a bit of a pain to haul a big mount and OTA but worth it (though I always question that during, but when I see the results and sure again). And to me remote imaging is not that different from just buying someone else's data -- interesting but very different than DIY. @Dale Penkala so this was intriging but did not work out that well for me. I used photoshop and did a color select to get the background and ignore the stars and small galaxies (and it did that reasonably well). But the issue is brightness not color for the splotchy portions. A complete desaturation made it more neutral, but not less splotchy. In the following the original is top, desaturation is the next. I can get rid of the splotchyness by reducing the brightness of the background however the selection can't distinguish the important fine detail (and my precious tidal tail) from the background well enough, so it is wiping out too much. I fear I am going to have a similar problem in PI with any selection tools. I also suspect, though not sure quite where, that my efforts at every step to preserve the tidal tail contributed to the splotchiness. And I feel like any area selection around it (or the other galaxy fine details) is going to leave too strong of a boundary even if fuzzed, a bit too artificial. I'm not a purist on using only global editing but do not want to "paint" the result either. More to do... I should note I'm a bit confused by everyone's focus on using NX in linear when the author says it works in stretched? Update: Sorry, forgot to attach image:  |
|
9.82
#...
·
|
|---|
|
For me, I use NoiseX or Deepsnr (results vary but I generally prefer NoiseX) at nearly every step in my processing. In the linear stage I use it on a mild level, usually 0.5 sometimes 0.6 (I leave detail at default 0.15). Once the image is stretched and curved, you'll usually bring out some more noise, to which I use another layer of it. Something that may help you... I'm one of those guys that stretches using AutoSTF. I change it up a little bit however. Once AutoSTF is applied, I drag the lower end down until all the noise disappears, and the upper end up until all bright cores, etc are visible. The rest of the stretching is done utilizing curves and masking. Utilizing this method I only curve the bottom end of the image until the noise is barely visible, then use NoiseX to curb it. Seeing your screenshots above is kinda surprising to me, normally NoiseX will take care of that. I'm wondering if maybe BlurX is tightening up the noisy pixels, and making things worse. Maybe try NoiseX before BlurX? Strange things going on IMO, but I can only speculate without actually getting my hands on the dataset. |
|
5.76
#...
·
|
|---|
|
So @Arun H that shows a lot of promise. I suspect I should have done it before downsampling also (it was drizzled, then downsampled back to the original scale). But what I ended up with when viewed relatively small is zero change, no suppression of visible tail or fine detail. When zoomed in strongly (2:1) I got what you see below. This followed your description, then used what seemed like pretty extreme ACDNR settings (std 8.3, amount 0.9, iterations 11, multiscale recursive and 5x5 mM with structure size 7, all picked more or less by trian and error). I then added NX at 0.58 and detail 0.25 to clean up some remaining small scale noise. I think this might be worth posting as a revision?  |
|
11.99
#...
·
·
1
like
|
|---|
Linwood Ferguson: Mostly I am glad it helped you. Whether you post it as a rev or not is your call - it is your image! Noise Reduction takes work. There is a tendency to think that one or the other tool is magic. It isn't. Noise X is the "in thing" now, but like any other tool, it has limitations and, unless careful, can be a blunt instrument just like anything else. Different tools have different strengths, and the skill is in deciding when and how to use these tools and not losing in the process the detail you worked so hard to capture. |
|
5.76
#...
·
·
1
like
|
|---|
Arun H:Linwood Ferguson: The trick of luminance extract and stretch for a range mask was very helpful as well. |
|
15.85
#...
·
|
|---|
Linwood Ferguson: Hello Linwood, Yes it will work in a stretched image but not as well as if it is in the linear state. https://www.rc-astro.com/resources/NoiseXTerminator/ |
|
15.85
#...
·
|
|---|
@Linwood Ferguson I took your JPEG image and did a little playing around with it. Hope your not made, but I did a few things that I had described, and I think it is better, but this could be better if the image was processed right from the start. I'll stand behind what I and several others here had mentioned. If you're willing, I'm happy to play with your integrated image without anything done to it and see what I could do with it, but that's up to you. Wilthout having the image from the start it's hard to say, but I think some of what you have is process induced. I've seen some of this before. While the magenta spots are still there, they are greatly reduced. |
|
5.76
#...
·
·
1
like
|
|---|
| Thank you Dale, not mad at all. The four stacks are about 7.5gb so not that easy to share, but I do appreciate the advice and effort. I plan to make another complete pass one day after I catch up on some other processing. |
0.00
#...
·
·
1
like
|
|---|
|
Hello Linwood, To me that is excellent. It is crazy loaded with galaxies and the amount of detail is amazing. Could it be improved upon, yes, but to what level?? That is nit picking, you did great. Just me, Joe |
3.58
#...
·
|
|---|
|
Nice dedication to put in those hrs. I've got the QHY600 so same chip and with that amount of hrs your background should be absolutely silky smooth. I would highly recommend giving luminance another chance and throwing most of your time at it and limiting each colour channel to a few hrs maybe 5 each. as you know doing the LRGB blend you can be really aggressive with noise reduction on the color channels so you dont need as much time. But that detail is really nice, well done |
|
5.76
#...
·
|
|---|
| Thank you Jay. |
