11.01
#...
·
·
2
likes
|
---|
I'm just wondering what processing techniques you might use to improve my 5hrs on M106? Please feel free to comment or even play with the image: M106 after 5 hrs integration |
1.20
#...
·
·
4
likes
|
---|
Frame it! |
0.90
#...
·
·
4
likes
|
---|
I would sit back, stare at it with a glass of wine in hand and tell myself "hey, I did that". |
7.75
#...
·
·
4
likes
|
---|
Jeremy Phillips FRAS: ... and based on the recent ap - life balance discussion, show it to your wife and enjoy the wine together ... Clear skies Wolfgang |
2.39
#...
·
·
2
likes
|
---|
I’d get at least another 10 hours of integration. Also the gradients are either from bad flats or light pollution but that is the next thing I’d address. Otherwise the star shape and brightness looks great and there is good detail in the DSO, it just needs more time and gradient correction. Good luck, I hope your skies are clearer than mine have been |
11.01
#...
·
·
1
like
|
---|
Blaine Gibby: They are actually from both; I changed so many things over 5 nights and even had my system somehow decide to change my camera offset for no apparent reason which threw my flats and darks all over the place. I'll see what I can do with DBE to mitigate some of that. The below shows you just how messed up the 5 days were: |
11.24
#...
·
·
14
likes
|
---|
Hi Andy, I don't know what tools or applications you use but in general practical terms, with a few references to common processes, here is what I did in about 5 minutes to your 5hr image. It was a bit rushed but you might be able to use bits here and there in your work flow. First I separated the stars from the object using Starnet++ in PI. Stars process: Remove blue color cast by de saturating blue or resetting the black point Check for and remove gradient Apply noise reduction to star mask Boost star color and brightness without affecting the background color by using a mask or manipulation of individual stars Check for new noise Object process: Remove green color cast by de saturating green or applying SCNR in PI remember to check for over saturation of magenta Check for and remove gradient Apply noise reduction to object image Additional mild stretch using GHS in PI or other levels / brightness controls Boost overall color saturation and vibrance of the object without over saturation of the background Apply mild sharpening Check for new noise reintegrate the object image with the starmask image Mild crop of the sides to standard 2:3 Hope this helps, CS's scott |
0.90
#...
·
|
---|
Andy Wray: Any stack? |
2.39
#...
·
|
---|
I changed so many things over 5 nights and even had my system somehow decide to change my camera offset for no apparent reason which threw my flats and darks all over the place. It looks like you need to make sure you are tightening your image train. With that amount of rotation it looks like it has loosened a bit. Or take flats for each night. you might be able to fix those gradients in photoshop. I don’t know anything about PI |
11.87
#...
·
·
2
likes
|
---|
@wsg that’s an excellent reprocess. Very nice getting the color diversity in the Galaxy. |
15.85
#...
·
·
2
likes
|
---|
Hello Andy, I gave your data a go and here is my result. Similar but slightly different. My workflow is as follows: I did a Dynamic background extraction, then did an Automatic background extraction. Used StarExterminator and then pulled luminance from that to sharpen a tad as well as bump up the contrast and add the natural color you captured. Depending on the look, I sometimes will give PCC a try too and compare to see what I like best, but I just used what you captured for the color. A note on the background of this image, I don't know if you got issues with your flats but when I got done even with the background extraction processes I had a hazy foggy look, so I went in and cloned out most of it. I didn't take a lot of time, but if I was to go thru and fix this right, I'd spend more time on this part to really make things smooth and blend in a bit better than I did. This can be a time-consuming process. I then used HDR Multi Transform & Local Histogram Equalization to pull more detail out of the core. Stars: I pulled luminance from your star mask and used LRGB Combine to add more color to the stars. Then I did a slight star reduction with Morphological Trams. Only about 35% btw Used pixel math to blend the stars back (about 90%) in to the galaxy. Once everything was added back together, I tweaked the background saturation by creating a range mask, inverted the mask and desaturated the background. Final color saturation and curves adjustment was done in Gimp. BTW I did no noise reduction because I used Topaz DeNoise to sharpen up the luminance layer. I also used Convolution to blur the starless image just a bit to smoothen everything out as well. This is far from perfect, but in the end I think it's got really good detail in the core and very nice natural color that you captured. More integration time would definitely help the outer range of the galaxy. Good Job! Dale |
0.90
#...
·
|
---|
Mirosław Stygar:Andy Wray: Nevermind. |
6.10
#...
·
·
1
like
|
---|
Hi Andy, I don't know what tools or applications you use but in general practical terms, with a few references to common processes, here is what I did in about 5 minutes to your 5hr image. It was a bit rushed but you might be able to use bits here and there in your work flow. Nicely done Scott, wow! |
1.51
#...
·
·
1
like
|
---|
Hi Andy, Looks to me that you've done a nice job with this image, noise reduction looks good (NoiseXterminator?) so I think any suggestions I offer would be "to taste" rather than any real technical improvement. Personally, I think I would experiment with two elements - color saturation and dark structure enhancement, probably by using Starnet to separate out the galaxy from the star field, playing with saturation a little (I like the way you have brought out the subtle blue in M106's arms), using the dark structure enhance script and then recombining the starfield back using pixel math. Looking at your image I'm pretty sure you know what you are doing anyway! I think that some of the other comments on this thread have been helpful, I thought Scott's approach of using a small star reduction with Morphological Transformation was a good one. So, if anything, minor tweaks maybe, but an image to be proud of! Paul |
11.01
#...
·
|
---|
Thanks for all the suggestions. I still have much to learn: I'm struggling with how to get rid of gradients and am also finding the stretching process from linear to non-linear difficult. Any suggestions would be appreciated. Here's my latest attempt. M106 after 5 hrs integration |
7.42
#...
·
|
---|
Just for starters, what have you used the remove the gradients and when in the preocessing pipeline? |
11.01
#...
·
|
---|
andrea tasselli: I usually use DBE in the linear stage. In this case I had a few problems as follows: * Lots of clouds * Moved from a normal guidescope to an OAG mid-capture ... that meant my optical train got rotated quite significantly to retain the 56mm backfocus * I didn't tighten the optical train enough when I installed the OAG, so 2/3rds of the way through I had to rotate it and had tilts issues for the previous 2 days * I didn't capture flats on the nights when I didn't realise I had a rotation problem * For some reason the camera driver decided to drop my offset from the default of 50 to 10 during the process ... I have never manually set the offset, so my darks didn't match. I had to use pixelmath to subtract the 40 offset off my master darks to try and compensate. I'm now manually setting the offset rather than leaving the driver to do it All-in-all: I'm having to rely on ABE/DBE and anything else I can think of to cater for my previous mistakes. |
7.42
#...
·
|
---|
Andy Wray: Let's assume you used DBE in this specific instance. What were the parameters you've used, if you can remember? |
11.01
#...
·
|
---|
andrea tasselli:Andy Wray: A tolerance of 3, and a smoothing factor of 0.6. Sample radius of 30, Maybe 7 samples per row. I would then move the samples to make sure they hadn't captured stars or were over faint areas of the galaxy and then manually add samples in areas of steep gradient I typically look for the worst cases of gradient out of each of the individual RGB channels; work on that and then save the DBE as a sample set (process) so that I can apply it to the other two. FWIW: I'd like to know the difference between subtraction and division as I see conflicting info on those two. |
7.42
#...
·
|
---|
Andy Wray: 3 is way too much. I wouldn't go beyond 1.5. Work on the minimum sample weight instead (e.g. from 0.75 -> 0.1). Smoothing factor is also too high. You need to go down not up, try using .125. Sample radius should be a more modest 15-20. Samples per row for APS-C sized sensors should be at least 66. Remove all samples overlaying on the galaxy including the halo. |
11.01
#...
·
·
1
like
|
---|
andrea tasselli:Andy Wray: Thank you for that; very helpful. I had a quick play as per below. I will, however, do a full re-process using your suggestions over the next few days. |
6.10
#...
·
|
---|
Hey Andy, this last version is really nice. Ian |
1.20
#...
·
·
2
likes
|
---|
Here's some tweaking on your latest... M106_HDR6SS32_HVLG_CBS_CBH_SS2083_Noise3020_USM8035_Curves_Sat15.jpg |