9.03
#...
·
·
2
likes
|
|---|
|
One more additional GHS stretch. Think I am getting the colors right or better now... if you stretch too much the NB gets light... signal moves into 3 and 4th quartiles. This one stretched a faint supernova reminant in O-III.  LBN 325 |
23.42
#...
·
·
1
like
|
|---|
|
Oddly this thread was "closed" recently for no apparent reason. This was an error and the thread has now been reopened - thank you Astrobin! Personally I'm pleased to see the thread reopened as I am loving seeing all the GHS related content people are posting! Clear skies everyone Mike |
 #...
·
·
1
like
|
|---|
Mike Cranfield: Yes, apologies for that. Either I mis-clicked or it's a rare bug somewhere on AstroBin. |
|
9.03
#...
·
·
1
like
|
|---|
Yes, thanks. I am really working with this procedure and it is very helpful. There are a lot of devils in the details here. Sh2 108 (Gamma Cygni Nebula) Have been working on the above a lot and this one has pretty rich texture again carefully stretching. It is a lot like Ansel Adams Zone technique for B&W where you try to evenly dill the histogram equally with all 11 zones of grey in his parlance. If you do that here the "white" is too bright for astrophotography... but I think equal distribution over the left half of the histogram is a goal... this one was weird and covered the whole histogram but did not register in Astrobin properly:  Pickerings came out more normal:  NGC 6979 Pickering's Triangle  |
|
23.42
#...
·
·
2
likes
|
|---|
|
Hi Jerry I am really enjoying your posts here - you have some great images, thanks for sharing them. I notice you are having problems with the blackpoint/linear stretch in GHS. Could I ask if you could elaborate a bit - there may be something I could do to improve things. I did notice that, when using the Low Clip slider on an individual colour channel, it is reporting the clipping proportion as though the black point is applied to all three colour channels. It would be more helpful if it reported based just on the channel you have selected. I will get that changed but, while I am at it, if there are other issues you have I could potentially address those at the same time. The histogram for SH2-108 is odd! Does it look like that when you view the image histogram on your computer, or is it only on Astrobin that it looks like that? Thanks Mike |
|
9.03
#...
·
·
2
likes
|
|---|
I think the histogram is some HTML issue since this is the Red channel in Photoshp: Pretty balanced: It is the image downloaded from Astrobin. Maybe that slider issue is part of my problem since most of my images are heavy in the Red/H-alpha. Other channels could be near null except for stars. I have new data from Sh2-79 where I collected more than usual. I have 4h with 450s images or 32 images and selected the best 19. I looked at 24 and there is H-alpha all over and I wanted to draw out the dim stuff, so this will be a good case to look at the details of the black point. I run this stuff in the background during the day. Here is that H-alpha data with just STF applied, will take you through the processing:  Plan is: StarXterminator NoiseXterminator GHS Here is StarXterminator... see noise a mile away now  Here is NoiseXterminator... noise gone now... magic...  Here is first GHS stretch with symetry point set to top of linear histogram:  Applied:  AutoZero Shadow Applied (No Pixels clipped):  Second Stretch. Symmetry point set between two humps... moderate stretch... some star removal artifacts... but a lot of detail...  Less "b"  Final Version Ha moved to R and RGB stars added in Photoshop. Note Red stretch in histogram cover left 3/4 quadrants.  Final version posted:  Sh2 79 GHS Processing in Detail |
7.75
#...
·
·
3
likes
|
|---|
|
Not 100% sure what you want to tell us. First comment: if you want to set the blackpoint you can achieve this within GHS by choosing the linear stretch. You don't have to use histogram transformation. Second comment: By placing your symmetry point between the two humps you increase the separation of the humps. If this is intentional, then fine. In case you want to get the two humps to overlap more, you could place a symmetry point on the right hump and restrict its extension to the right by highlight protection. I find it very useful to look on the logarithmic histogram on occasion to check for the highlights and the shadow regions. Clear skies Wolfgang |
|
23.42
#...
·
|
|---|
First comment: if you want to set the blackpoint you can achieve this within GHS by choosing the linear stretch. You don't have to use histogram transformation. To amplify Wolfgang's point - in case anyone is not aware, by clicking the button highlighted in red you can auto zero the clipping exactly like Jerry has done in HT. The corresponding button next to the HCP slider will do the same at the top end. Alternatively if there is a percentage of pixels you wish to clip then enter the number in the Low Clip edit box (eg to clip 1% of pixels enter 0.01) and BP will update automatically to reflect this.  |
2.11
#...
·
·
1
like
|
|---|
|
Hi Mike, first let me say again that you and Dave developed an unbeatable function. Before GHS, the stretching phase was always to me a source of frustration, and now with GHS it is a pleasure, thank to the precise contrast control you built-in. After being using GHS for a while, I have a user interface requirement. When tuning the stretching parameters I use to go back and forth several times from the main script window to the image inspector, to review details (e.g. Star cores). While the image inspector is open, the parameters cannot be updated, so I have to cycle into this openImageInspector-view-close-update-reopen loop, that is a bit time consuming. I would be happy in many cases just in having a bigger preview window, but Preview is not resizable. It would be great to integrate the image inspector function in the main window or as an alternative just having preview resizable. Thanks again for your great creation! Mau |
|
23.42
#...
·
·
2
likes
|
|---|
|
Hi Mau Thanks for your feedback. I am pleased you are finding GHS useful. As regards your request, the preview is already resizable. If you go into the preferences dialog ( the spanner icon at bottom right ) you will see the ability to resize the preview. I hope this helps but please let me know if I have misunderstood what you mean. CS Mike |
|
9.03
#...
·
·
1
like
|
|---|
|
Appreciate the comments and alternative methods. My issues are: 1. It is not intuitive for me to use the "Shadow Protection Point" in the script. There seems to be nothing in the shadows, if you use the noise reduction first. There has been discussion to be careful with the noise reduction as there may be signal in the noise, but if there is in my case, it is very very low. Thus I find it easier to zero the black point using one of the other methods. 2. The double hump after the initial stretch. I think that it may have been mentioned in the video that one hypothesis explaining the double hump is that the background is the left hump and the object is the right one.... and of course the transition is messy.... but if this is correct, then the best place to put the symmetry point for the second stretch should be between the humps to enhance separation of background and object. I just did another Sh2 (78) this AM the same as 79 and the double hump is there too. If this in some way reflected reality, it might be an efficient way to separate the background from the signal... Maybe I have over-simplified the matter, but these are my thoughts. |
|
23.42
#...
·
·
1
like
|
|---|
|
Hi Jerry. I think there may be some confusion here between the Shadow Protection Point and the Blackpoint. The former (Shadow Protection Point) is one of the main GHS parameters and defines a point below which the transformation function is linear. It tends to “push” the transformation graph right, ie it prevents the dimmer parts of the image becoming too dark. The latter (Blackpoint) sets everything to the left to zero - this is the equivalent of the HT Shadows parameter, confusing I grant you! The Blackpoint parameter in the GHS script is accessed via the linear stretch transformation type. I hope this clarifies a bit. incidentally I did PM you with some additional info. All the best Mike |
|
2.11
#...
·
·
1
like
|
|---|
Mike Cranfield: Thank you Mike for your swift answer. I missed that the Preview size control was in Preferences! And, yes this addresses my question. Many thanks and CS! Mau |
10.10
#...
·
·
1
like
|
|---|
Appreciate the comments and alternative methods. Hi Jerry, As Mike says, the black point is very different function than the LP parameter in GHS. In GHS, there is no clipping. Rather, below LP, the stretch is linear rather than the GHS function to the right. I suggest you load up and image and set up to do a moderate stretch on an already non-linear image to see how the transform is changed by the use of LP (and HP for that matter). You can see that both LP and HP can be used to control the overall brightness of the stretched image. The double hump is and indication that you may have put too much contrast over a short interval (too much, too highly focussed) and does not generally yeild pleasing results. If you start to get a double hump you should consider backing off b and/or D. Alternatively you can place SP closer to the peak of the histogram. In general, the best place to put SP is in the "hills" of the histogram as this is likely where contrast is needed (the pixels are bunched up here). Valleys are places where few pixels exist an generally you dont want to add contrast here. Finally, it is personal preference, but I get a lot more out of the log histogram than the "linear histogram" view - especially after the initial stretch. This is where I hunt for hills to place more contrast. Hope this helps, Dave |
|
9.03
#...
·
·
1
like
|
|---|
|
This is all very helpful for a very complex process, but it seems you have provided all the tools and I just need to try to use them intelligently (if that is possible). I read you on the double hump and have backed off on this new image (Sh2 110)... guess better to do 2-3 small steps than 1 big one. Here are a couple variants of the LP (on a linear image). Seems the more linear component, the more enhanced the dark side of the histogram?      |
|
10.10
#...
·
|
|---|
|
Jerry, Thanks for your posting, your series illustrates the effect well. I generally haven't used LP or HP on the first stretch but nothing to say that that is not good. Cheers, Dave |
|
9.03
#...
·
·
1
like
|
|---|
|
Thing is that if you use NoiseXterminator or other serious noise reduction on the linear image as I have been taught, there is very little in the dark side of the histogram (hopefully not much is lost). Normally I also remove stars but I have to be careful with the stretching because StarXterminator, which I think is better than the com petition, still leaves artifacts that show up in the stretched image... along with the target detail... these are normally taken out in Photoshop with the repair tool... again as I have been taught by various gurus. Normally I remove the stars first with the hope that the noise reduction will clean things up a bit. |
|
9.03
#...
·
·
1
like
|
|---|
Well, one more adventure in GHS: Sh2 78 This was very interesting to me. I toned down the stretch a bit and used some of the tools mentioned to deal with the darker side. In the image I put up a more heavily stretched Ha version that picks up more structure but also more noise. This thing looks square when you look at it but is supposed to be an old PN. I think there is some circular structure thee on the stretch, but hard to see how to process unless I use a mask.  |
|
9.03
#...
·
·
1
like
|
|---|
Yet another.  Sh2 110 Fine tuning the stretch to cover the range of Red values... also manually remove start artifacts in Photoshop...  |
|
10.10
#...
·
|
|---|
|
HI Jerry, Just a couple of comments. Is that noise in your RHS image or IFN. I kind of like it. I think this subject would also benefit from some O data, and I think it would make it overall more round. Nice image though. I still like the monochrome Ha one. For emission nebula, Ha just dominates the Green and Blue signals. Dave |
|
9.03
#...
·
·
1
like
|
|---|
David Payne: Surrounding Sh2 78 is pretty much noise except for in close, I think... this is heavily stretched. Yes, none of these have any O-III and there is a lot to be said for seeing the detains in greyscale. I have a few LBNs in my list that will have a wider pallet. FYI, I did a simple stretch with GHS and put that up for Sh2 110. Any more stretch and the stars bloat..... why I destar and add back RGB unstretched stars. https://www.astrobin.com/ioh1av/#rB |
|
9.03
#...
·
·
1
like
|
|---|
Well, I am still working away at this. This object has strong NB signals in all channels and I destarred, denoised and GHS stretched each individually, then equalized the histograms.  NGC 6914 If you are interested, this is the combination recipe: Dynamic Combo |
|
9.03
#...
·
·
1
like
|
|---|
Guess I cannot stop working on the fine points of GHS. This one worked on highlighting these filamentous threads. In addition to the usual careful stretch I used some Photoshop techniques to use a HighPass filter and masks on the target areas. There is a lot with which to work with the combination of the CMOS and GHS. Sh2 113-114 Supernova Reminant? |
|
9.03
#...
·
|
|---|
Well, still plugging away. This is using GHS plus the Dynamic Combination Pixelmath to crank up the O-III on the Crescent Nebula. This was tricky with GHS because the Combination technique cranks up the O-III signal such that it does not overlap well with the Red/Ha/S-II signal. Have to watch where the symmetry point lands.  Crescent Nebula O-III Emphasis |
|
9.03
#...
·
·
1
like
|
|---|
|
This one has all three NB stretched by GHS and integrated. Individual NB frames presented. Previous image three NB integrated, then stretched, this is three NB stretches and one integration. I think more control. North American Nebula... barely fits  Sh2 117 Dynamic Narrowband Combination |
