Hi I like to keep stars as a separate layer in photoshop during my final edits (after PI processing). What I usually do i use StarX to create a star only image in linear, stretch the star only image and blend in PS over the processed starless with "screen" blending mode. What is the benefit of extracting stars from a non linear vs linear+stretch star only? Is my method equally valid for blending stars back in? Thanks!
Let me carify some more :-) 1) This stript was designed for people who want to remove stars using the best matimatical method for both "linear" and "non-linear" data. When removing stars on a non-linear image, the best way to do this is to return the image to a pseudo linear state, then remove the stars, then return it back to its original non-linear state. We call this method "Reverse Stretch". By doing so help retain star luminance & colors when transfering them to another starless image. When we use this script to "add" the stars back to another starless (modified) image, we perform this "Reverse Stretch" method again using our script. Again, this helps retain star luminance & color. 2) This script was also designed for people who dont use Star Xterminator (SXT) i.e. Starnet++ and Starnet 2. Example, if someone did not buy SXT and used Starnett++, how do they properly remove the stars?? The substraction method used in Starnett++ is not the right way to do things unless the data is linear but Starnett++ works in non-linear mode so this is a problem. Also, even if you own SXT, our method of removing stars per(1) above is the better way of doing things, let me explaing more below. 3) The proper math to remove stars via "unscreen" mathmatics is (Starry-Starless)/(~Starless). The issue with this math is you can get errors when dividing by zeros. This becomes an issue if you have an image with a blown out core, like M42 the Orion nebula which was showing 1 values for the areas blown out (the inverse of 1 is 0), so you would get errors in the stars only image because Pixinsight would return the divide by zero error into a 1, so now your stars only image would be corrupted in areas which were blown out. SXT uses this formula for "unscreening" to avoid dividing by zeros (Starry - Starless)/(1.005 - Starless) which they create an offset to the data, but SXT also says to not use "unscreen" process for star removal on linear data, because a) the 0.005 offset they use in the formula to avoid dividing by "zero" could make for a big change or disortion when later processed; b) if you dont use the inverse of this formula to add the stars back, then you are working with apples & oranges and things may not match properly. So SXT says on linear data to just remove the stars with "Unscreen" option unchecked to avoid issues, which uses subtraction instead. In pixinsight, the easiest way to manage unscreening stars is to use this formula "rescale(Starry,Starless,1)" which will manage the divide by zero issue for us. So because this divide by zero issue is fixed, in our script we can benefit from this on both linear and non-linear data, thus we can control the exact methods of star removal the same way as when we add them back in as long as you use our script for both process. I also want to make something clear to everyone who reads this comment. SXT is a very good process for removing stars from an image and is probably the best. It is because of this starless image SXT can create a stars only image by using the original starry image and using subtraction method for linear data (unscreen box unchecked) and unscreening math (unscreen option checked) depending on what option you pick per the data type, which is why Russel, whom I consider a friend, offers these options. But people may not know what option is best to use and they might be unscreening linear data when they should not be. The thing is, a lot of people also process in non-linear mode, because in non-linear mode we can do things like star reduction or narrowband normalization, Ligtness management, etc. which you cannot do in linear state. Times are changing and processing images in a non-linear state is becoming more common. So, when removing stars from a non linear image, we can still use SXT for the starless image only, but then use our script to properly handle the "Stars only" image part as our method of doing this is more controlled as described above, espcially when we screen the stars back to a new starless image. If you were to test this on an image with heavy nebulosity, you will see the difference. So in your case, if you are only using SXT and the star only image it creates, in a linear state, and you process this in PhotoShop, this is perfectly ok as long as your "unscreen" check box is unchecked in SXT. However, we still think it is better to remove the stars on linear data using our script; Example. What if your "linear image" is over exposed i.e. say you imaged M42 at F2 with longer exposures and the nebulosity can be clearly seen not streetched. STX says to not unscreen if the data is linear, but in this case I would because of how bright the data is. But even if you did, you would get better results using or script because even though it is "linear", we have the option to reverse stretch it thus treat it as nonlinear, do the reverse stretch re (1) above, and give you better star output especially in areas of heavy nebulosity (as shown in the video). And for people who do not use SXT for the starless creation, or for people who work in non-linear mode, our script is what you would want to use to remove and add the stars. Its all about controlling the workflow and I guess I need to produce another video to detail this better. I hope all of this makes sense now. :-)
Please read this write up that Mike Cransfield did on subject on the Pixinsight forum to help explain things better: pixinsight.com/forum/index.php?threads%2Fnew-script-screenstars.21098%2F
@@anotherastrochannel2173 many thanks for the comprehensive reply! And btw impressive work with the script. Quoting Mike's comment in the PI forum "For ScreenStars, therefore we use the unscreening formula for both linear and non-linear data. And, in keeping with Objective 1, adding stars back is always done by the conventional screening formula whether linear or not.", I conclude that for my workflow it would be best to use the scrip (instead of STX) to extract the stars from the linear image, then follow my normal routine of stretching the stars only image in PI and screen blend it as a top layer in PS. The "Reverse stretch" method is only used for processing already stretched images.
Hi guys All of your tools are amazing and I hope that everyone is watching this video until the very end and do a donation to encourage and support you guys. You deserve it! Many thanks Bert
Your work is brilliant, Bill! I am eager to try out your star removal method because in AP we often see stars as something to get out of the way and overlook the beauty they have in their own right.
I agree, sometimes I like the star image better than the rest. The brighter the target is, the harder it is from a color stand point. The reverse screen helps with this but like everything else in life, nothings perfect, but it does help.
Again, I am suitably impressed with your wizardry. I was able to control bloated OSC stars that were taken with a UV/IR Cut filter of M42. One thing that I found was using Linear images, I was unable to select the "Enable reverse Stretch" using Version 1.1. Thanks to both you and Mike.
Fantastic work, many thanks Bill and Mike. I’ve just used this on the Elephant Trunk and North America Nebulae and it’s worked great to bring out the star colours and reduce the brighter ones to really show the Nebula detail. Clear skies Ian
Awesome - I love that this is a script! ...and brilliant. Thank you for your continued contributions to astro photography processing - also, great image of the Lagoon and Triffid
I'm having fun with your script and I see a difference in the saturation of the star colors with your ScreenStars script. I compared it with your PixelMath expression " RGBstars to NB V11 ". I think I'll remove this PixelMath expression from my list of icons set on my narrow band workspace. Thanks for the script Bill I’ll support you from your donation link for sure.
Christian, actually, "for" true color replacement, the RGB stars to NB pixelmath script is the best way. In fact I recently updated the math (V14) but this time I am converting everything to a CIELAB color space because the lightness control is better then using luminosity. We plan on making a script for this too. As for the Screen Stars script, it is a better way to transfer stars over, but if the new target brightness is to bright or too extreme of a color change, you may see some discoloration still, but its better than using the conventional screening method. Hope this makes sense.
I see that this narrative is an overview of v.1.0. My copy, v.1.1, no longer includes reverse stretch option. Nevertheless, this script is an easy and painless way to achieve effective star reduction. Thanks so much, guys!.
Great tool! I was wondering if you could add an enhancement to it to remove small stars. I know there are tools that reduce star size, but then you loose a lot of good looking stars. I have images where I want to keep the stars but not the star dust if you will. Very nice script!
This is amazing, thanks so much for the script and the video. just to clarify, the image with stars you loaded at the beginning of the process - was all stretching and editing done to the nebula with the stars in? should i remove the stars at the beginning then carry out the stretching/curves etc or do i do the whole process with the stars in the image, then do the screen script? thanks again for your help, love your channel
I actually use either SXT or Starnet2. Both do a great job but it all depends on the target. The best part of SXT is that it is currently supported and continues to improve, whereas StarNet2, I'm unsure of. Both give bot a try. Regular Starnet++ leaves artifacts so I don't use that at all anymore.
@@anotherastrochannel2173 Thanks. I am curious about the difference between the UnScreening of stars that takes place in STX vs the StarRemoval of your script? Thank you so much for all you have worked through.
Even when you are not using the reverse stretch option - ie just "conventional" unscreening - there is a small difference between the ScreenStars script and the RC-Astro SXT unscreened stars. In SXT there is a small adjustment introduced designed to compensate, in a rather approximate way, if your starless image has significant bright clipped pixels (i.e. blown out nebulosity). It does mean, for example, that if you combine your starless image with the SXT unscreened stars using the familiar screening formula, you won't gat back to where you started! ScreenStars uses a "true" unscreen formula, i.e. the precise inverse of screen blending. If your data is in reasonable shape, using ScreenStars is definitely preferable. If it's not in reasonable shape, it's best to go back and get it in reasonable shape and then use ScreenStars!
I'm assuming you're using StarNet or Star Xterminator to create your initial starless images. Each of those processes creates (or has the option to create) the "stars" image. It would be nice to show the differences in what your script does when creating the "stars" image vs what those processes do by default. I think you kind of hint at it in the beginning of this video when you show how the "reverse stretch" option works. This is definitely a neat little tool, and I'm going to grab it to play with it... see how it compares to my typical processing workflows. Not like I'm able to actually image anything with the absurd weather we've been having... so might as well play with some old data LOL!
Jonny, so Starnet subtracts the stars, this is bad unless you are working in linear. SXT using a regular unscreen method. Our method we are performing a reverse stretch of the non-linear data, then unscreening, the returning back to non-linear which preserves the star detail in areas of heavy nebulousity. The comparison I did in the video between regular unscreening and also the reverse stretch method will be the same as our method vs SXT starless image. Play around with it for sure, and I completely understand about the weather situation. We finally got a break in the past 2 weeks for decent weather, before that it has been horrible. Its actually clear tonight, but too tired to do anyhting, lol.
Dear Bill I am sorry but I don't understand the basic thing: Why do I need a starless image (RGB_starless) when I want a "star only" image from that RGB image? The RGB_starless image was made by another star removal process like SXT. So when I have only a starry RGB image and want to get a star only image from that, why do I have to select a starless image from the starry image? Sorry again for not getting the point! :) So your script is not an alternative to SXT resulting in a starless and a stars only image? But looking forward to the other scripts and appreciate your ongoing effort in releasing awesome scripts! Thanks a lot, Nico
Because starry image minus starless image equals stars only image. but instead of "minus" we unscreen the starry image from the starless to get the stars only, but to improve this even further, we perform a reverse stretch on the two data sets to a pseudo linear state, then unscreen, then stretch back, to give us a stars only image. This method produces a better stars only image. When we do this same reverse stretch screening process to transfer the stars over onto a new starless image, we get better star color protection. Re SXT, SXT is only unscreening the stars from the starry and starless image to give you stars only image (behind the scenes), we are doing it better by offering a reverse stretch. Please compare to see the difference. I actually showed the difference in this video from conventional unscreening and performing a reverse stretch unscreening, you will see the same thing. Again, in the video I expressed the big "difference" in both methods will be seen in areas of heavy nebulousity.
Here is the data I used in the video if you want to compare as I did :-) : drive.google.com/file/d/1Tz-XZ5jlI1s9TEhTx_f6FSfgHNwNeRvU/view?usp=drive_link
Dear Bill Thanks for your clear explanation and the data! Of course, now I got it!🙈☺️ And I will include it in my workflow - but will still use - as I read - your other script in case of inserting rgb stars in a nb image. One last thing: I often struggle stretching a linear stars only image just to the right amount regarding star size, brightness and halo radius/intensity so that the stars match the separately starless image afterwards. I wonder if there might be a pixel math stretch process in future... ☺️🙏
1) We are removing the stars in a better mathimatical way. 2) There are limitations to using Star Xterminator (SXT) when it produces a star only image; example: sometimes SXT leaves part of a star on the starless image which has to be cleaned up, the problem becomes, how do you get part of that star back onto the star only image?? The answer is to not use he SXT star only image, clean up the staless image by removing the star remant, then use our script to properly produce your stars only image. Does this make sense now?
@@anotherastrochannel2173 I think so, so in steps, 1. Create a clone of the image 2. SXT the clone, dump the star only. 3. Clean up the Starless image, 4. Use your script to produce new stars only image?
I've read in a few places that you might remove stars while in linear state, then process background and stars separatedly, and close to the end place the stars back. What I could not find is any tutorial/advise/workflow for processing the isolated stars. Anyone can help?
Why do i see "view geometry differs" on the preview screen of the script instead of an actual preview? Also the script fails to run. I star aligned the "stars" image and the "nebula" image.
Hello Bill, I wanted to know how to remove the stars using the star screen script in the first place because when I opened the scripted It is a requirement to have a starless image already. Am I missing something or should I use starXterminator to remove the stars? Thx for the great tools you are providing..
Yes, use SXT or Starnet2 to remove your stars to give you a starless image. You dont need to use the SXT "remove stars" feature, you will do this in our script as the way we do it using the "reverse stretch" math is a better way. Watch your new Star Reduction Script video which I just released to see how they both work together.
hello I am using Screen Stars Script on Mac and Enable Reverse stretch box is grey. can not tick it . please help what I am doing wrong. many thanks Krzy
Are you working on linear data? If the script detects that the image has not been stretched, or only very lightly stretched, it will not try to do a reverse stretch and the option will be greyed out.
Is it possible to modify this script to allow rescreening of mono stars onto a color image? Sometimes mono stars are used for artistic effect but the tool in its current configuration doesn't like that combo.
I took a stretched OSC image, did a star removal, and then star replacement (both with reverse stretch enabled). And I created a star image using Starnet2 and did a simple starless+stars Pixelmath. And both results are identical. So am I missing anything or whats the advantage or usecase of the Script?
In the video I said that both images will look the same in areas of no nebulosity, but in areas of heavy nebulosity, you will see the difference, as I shown in this video. Test this on some data with heavy nebulosity. If you dont have any data, I can send you a link to test on 😀
I am seeing the same thing happen. Added the repository, checked for updates, installs the script, restart PixInsight, but no "Screen Stars" appears under Utilities
Hi Bill, Thank you for your continuing help to the astrophotography community. I have followed your instructions to install this script and in the Installed Updates page of Manage Repositories it shows up but I cannot find it in the Utilities drop down..... I notice that in the installed list the script shows as a single line with a scroll type icon whereas with the GHS script for example it also has other lines one with a gear icon. I wonder am I missing something? I have tried to repeat the installation with no luck. Any help to a PI newbie would be most welcome. Thanks again
In the installed updates page the scroll icon signifies a script, the gear icon signifies a process. GHS exists as both a script and a process, hence multiple lines (also a documentation line). For Star reduction you should have a scroll and a document line. I don't know why it is not appearing in Script>Utilities menu. Hopefully the files are there and you can install manually via the Feature Scripts utility - assuming you know how to do this?
Thank you for the replies gentlemen. I have successfully installed the Star reduction script with no problems at all. In the Installed updates list for Screen Stars there is only the one line, no documentation line as with Star reduction, is this correct please? The address I used was copied as your demonstration Bill and when I look in the Feature Scripts list Screen Stars is not present Mike. Sorry to be a pain, thank you.
@@petecarpenter1068 It is correct that there is no documentation line for Screen Stars - we did not feel additional documentation beyond the comprehensive tooltips was necessary. I don't know why Screen Stars is not in the Feature Scripts list - one option would be to find the updates.xri file in your PixInsight directory and rename this (eg as updates.xri.old) then restart PixInsight - it should then reload all of your PI updates including the official PI ones as well as Star Reduction and Screen Stars.
How the heck do you enhance the dust in the background. Even with a lot of hours, my background just shows noise, and I can never get dust, just removes the noise and I can't figure it Out.
I'm not sure what your setup is, but this was about 4 hours shot from bortle 8 on a 80% moon lite sky using a Baader Moon and Sky Glow filter, Edge11 with F1.9 Hyperstar using the PlayerOne Poseidon OSC camera. Make sure you shoot broadband in order to see the dust 😀
@@anotherastrochannel2173 ok, then I need to learn how to combine narrowband OSC data with broadband to pull that off. Then maybe range selection to seperate the narrowband object and blend it to broadband background. . Thanks!
Sorry but I'm even more confused than I was before watching the video. Why do I need a starry image and an image of just stars to remove them from the starry image. I must have done that already ?
You only need a starry image and a starless image to remove the stars. To add stars to another image, you will need the stars only image and the new starless image. You might want to watch the video again and then you will see what we are doing 😀
at 04:35 "I already created a starless image". How? With this script, some other script? I want to remove stars from an image with foreground galaxy and background stars. I'm looking at the script GUI and I see in the files section: Starry View, Starless View and Stars view. Can I, and if can, how exactly can I load my xsif file of M106 and background stars in it, remove the stars from it, generate a starless view of the galaxy and a "Stars View" which I assume is only stars? Sorry, I really appreciate your work, but the whole reason I installed this script in PI was to use it. not starnet2 or other star removal plugin, to remove stars from existing image and generate the equivalent of a star mask and a starless image. That information is absent in this tutorial.
I dont see the point in using SXT stars when our method of star removal is a better method. However, if you dont want to use our star removal method and just use SXT stars to add to a starless image, you can use our ScreenStars script "Star Replacement" option and dont use the reverse stretch option, as this uses is simple screening math.
I watched the video again, and realized that this tool might be only for OSC (one shot color) camera data. I shoot mono with filters, so would this tool be of any use for me? If so, then what are the workflow steps for mono data?
Hi, thank you for all the work you've done. I've read through all the comments here, and I, too, am totally confused. So your script needs both a starless image and the original with stars. How exactly are you supposed to get a starless image without first removing the stars from the original image? Isn't that what the title and description of your tool is, "This script provides a tool for extracting and recombining stars." ? Also, what is meant exactly by "screening" and "unscreening?" I see those terms thrown around a lot on various videos with no explanation. If you have to use another tool to extract the stars from a complete image, then what is the point of your tool? Please explain to a total dummy. (I think there are a WHOLE LOT of people out there like me earnestly trying to learn the deeper levels of Pixinsight). BTW, I enjoy learning Pixinsight, but sometimes the frustration gets the better of me. I do SHO and RGB with a monochrome astro camera, with mainly cookbook type workflow gathered from TH-cam sources like yours, constantly evolving as I learn more. Thanks!
You create the starless image using StarNet2 or Star Xterminator, then you run our script. Once you have the starless image, you can use our Script to properly remove the stars. You do not want to use Starnet2 or SXT for removing stars as our script does a better job of it. So after you use our script for removing the stars, you can later use our sctipt to add the stars back to your modified starless image.
I guess I need to do a better job of making videos. I will do a follow up video on this next week. I do suggest you watch some of my other videos, like adding RGB stars to Narrowband, etc, as I do cover some of the math re adding/substracting vs unscreening and screening, this might help you understand.
@@anotherastrochannel2173 As an engineer with 25+ years background in aerospace and backend database software, I can certainly understand the old adage, "the more you know, the more you realize how much you *don't* know." I always had a vague knowledge of image processing, using basic tools to edit regular photos, like contrast, brightness, hue, etc., but never really knew how any of that worked at a software coding level. Coming into Pixinsight with that little bit of knowledge about a year ago (also minimal Photoshop experience at the most rudimentary level), it seems to be very much geared to those with an expert level of knowledge of image processing. I think there are multitudes of people out there like me, who wanted initially to apply a simple cookbook recipe approach to processing astro photos (I jumped straight into narrowband/monochrome without going through OSC first), with workflows gleened from Pixinsight TH-camrs like yourself. I can do pretty well and make impressive final images, but I don't really know why or how I am doing it--I'm just following others' recipes and tweaking the processes for my own data for the most part. At some point, I wanted to gain a much deeper insight into what I was actually doing with all these tools. That's where my issue arises. Pretty much all these TH-cam tutorials assume you are either a total newbie with no knowledge of image processing (and they don't give you any detailed explanation), or you are a total expert (and they don't give you any detailed explanation). There seems to be no material geared toward the intermediate level user. It can be very frustrating to watch tens of hours of TH-cam videos just to get that 30 second nugget of knowledge that leads to that "Ah ha!" moment when it all comes together. I think I speak for many tens of thousands of astrophotographers when I say that. Concerning your "star removal tool," my take on it is that you are taking advantage of another tool already in common use that produces a starless image, which you then backwards engineer somehow with the original image to get a "better" stars-only image. I think the main question/confusion that people are having (like me) is, why don't you just make your tool so that it extracts the stars in the manner you want from the original image, without needing yet another tool to create a starless image? This explanation would be extremely beneficial to an engineering-minded person, which is what Pixinsight seems to be all about. After experimenting with Starnet and SXT, it's obvious that those tools don't just "remove stars," since it would leave darks spots on the starless image where the stars used to be. It actually fills in those holes with some sort of AI voodoo magic process--not perfectly though, because on my data, it sometimes leaves little (and not so little) puffs of brightness from the halos, which I then have to use clonestamp to brush out. This has led me to much experimentation on *when* and *how* to apply certain tools in the workflow, and it all makes a huge difference in the quality of the starless image. A huge difference. The reasons for the order in which things need to be done are never really explained well in most videos, and I have wasted many days trying to figure which process order gives the best results. Another issue/question I have with star removal is the inability to run any kind of SPCC or PCC on the linear, combined colored (SHO or RGB) starless data in order to correct the colors. These tools cannot run without stars in the image, as far as I can tell. So, what is the point of SPCC if you can't run it on a starless image? It seems like an impossible situation. And it only works on linear data. These are the sorts of things I run into almost every time I try to learn and add a new process or script into my workflow, resulting in a complete upset of the order in which I do everything. I might have 15 or 20 different steps, depending on the object, quality of data, etc. It often seems as if there is no rhyme or reason to what some Pixinsight TH-camrs are doing compared to other Pixinsight TH-camrs. For someone who is at an intermediate level, and unable to really explain what I am doing at a fundamental level, it can all be extremely confusing and confounding. I simply am not able to make any judgement as to which approach is correct, other than many, many hours of trial and error. That is not very fun. Another killer is not knowing which processes work only with linear data, and which processes only work with stretched data. That always gets me, and I have to constantly return to the drawing board with my entire workflow, reinventing the wheel over and over. The most helpful video would be to watch an "expert" take some image data that he had never seen before, and do a processing video without rehearsing anything. Watch what he does when problems arise. That's where 90% of my learning comes from. We need some sort of apprenticeship program for Pixinsight. It's the most unintuitive pursuit I have ever attempted in my life (other than learning UNIX and getting along with women). Thanks for all your work, and I certainly understand that people are taking a risk putting themselves out there on TH-cam as a useful service and get all sorts of guff. But I just wanted to reflect what at least tens of thousand of people are thinking about Pixinsight and learning its various tools and methods. Cheers!
Hey Bill. I am using a mac and when I open the script Iand enter the two views, I don't get anything in the preview window. Any suggestions on what I am doing wrong? Thanks
You lost me, how did you remove the stars in the original image? I have a starry NB image and I need to remove the stars from it and later replace them with RGB stars. How do I do that? When I put that image into the script and select "star removal", and hit the check, it just fails. What did I miss?
Its all in the video. To remove stars from an RGB image, select star removal option, load your RGB starry and RGB starless image, then remove the stars. use the reverse stretch option too if your image is non-linear; this will create your "Stars Only" image. To add the RGB only stars image to your new narrowband image, select the star replacement option, load your stars only and narrowband starless image, use reverse stretch if data is non-linear, then create the image. Note: Make sure your images have been registered at the beginning of the work flow so they match image size and have propery star placement.
@@anotherastrochannel2173No, actually it's not. This confused me as well, but I think I know what you did. What he is asking, and what you didn't show, is how you got an RGB starless image to use with this process. I believe you used SXT to generate the starless image and you toss the Stars image that SXT produces. You then use the starless image that SXT produces. I'm realizing that this processes isn't a replacement for SXT or StarNet for the purposes generating a starless image, rather it's a replacement for those two processes for the purpose of producing better stars to add back into your starless image. It's an awesome process, and absolutely amazing contribution to the community, but it could use better explaining, like from a workflow perspective. I envision using this immediately after my initial stretch after (graXpert, SPCC, BlurX) and at the very end when I need to add stars back in, followed by star reduction.
I tried to use your script.I have two non linear images of the same subject, one with 2 hours exposure and the other one with 16 hours of exposure time. From the first one i want to use the stars and put them on the 16hours starless image. I got the star image in reverse stretch mode and it looks good, but it does not let me replace the stars with the script. apparently i'm doing something wrong. Maybe somebody can tell me what i am missing?
Did you register both images together so they match in size? If the images are not the same size then it won't work so you need to make sure that you register one image to the other
Hi there. I am a bit confused. At the Linear example you adding stars to a linear image which has been altered with NBNormalization. I thought that NBNorm should be used on stretched images. What am I missig. I am using StarXTerm at the linear stage and putting them back after stretching, but I have troubles to stretch the "OnlyStars" image properly without blowing them out. STF+HistTransf is very tricky for onlystars. What tool do you use for stretching onlystars images? Thanks a lot
Tom, You cannot use STF to histogram transformation on a stars only image because the STF is stretching an image to a known median background value thus it blows the stars out. So if you're going to use STF then you need to reduce the mid-tones bar so the stretch isn't so extreme, otherwise don't use STF and just use histogram transformation to stretch your stars. I personally do not do any processing in linear stage because of issues like this. There are many advantages to processing a non-linear using narrowband normalization but one of the reasons why we did the screen stars script is so that you can remove the stars from a nonlinear image and use the reverse stretch option to remove and add stars back to an image this have all the advantages of removing and adding stars from a pseudo linear stage perspective but doing it on a nonlinear image.
Much Thanks Bill for replaying to me, you are doing hell o a job, crating new things and explaining them both.. I am still learning, RC-Astro recomends to remove the stars as soon as possible (before stretching) so I was bit confused.@@billblanshan3021
If you’re doing it non-linear why is there a need to remove and re-add them if you stretched the image with them? It looks like you have 2 stretched images and you just removed and then added them back? Please tell me what I’m clearly missing haha. I understand doing it when it’s linear because when you remove the stars, you can get a better stretch on the nebula without having the stars getting blown out and what not but if you’ve already stretched the data, why are you removing them and adding back in? Clearly, there’s an advantage I’m just missing what it is. Is it strictly for putting RGB stars on a non-RGB image? In the first example it look like to RGB images so that’s why I didn’t make sense, but then I saw later are you adding RGB stars to an SHO image?
@@billblanshan3021 would you recommend removing them before stretching in certain cases to get a better stretch or do you always stretch them with the stars still there?
@@rvoykin You really have a choice, you work in linear phase or in non-linar phase. I would not mix. Wait for my next video, you will get the idea of where we are going with this 🙂
Hi I like to keep stars as a separate layer in photoshop during my final edits (after PI processing). What I usually do i use StarX to create a star only image in linear, stretch the star only image and blend in PS over the processed starless with "screen" blending mode. What is the benefit of extracting stars from a non linear vs linear+stretch star only? Is my method equally valid for blending stars back in? Thanks!
Let me carify some more :-)
1) This stript was designed for people who want to remove stars using the best matimatical method for both "linear" and "non-linear" data. When removing stars on a non-linear image, the best way to do this is to return the image to a pseudo linear state, then remove the stars, then return it back to its original non-linear state. We call this method "Reverse Stretch". By doing so help retain star luminance & colors when transfering them to another starless image. When we use this script to "add" the stars back to another starless (modified) image, we perform this "Reverse Stretch" method again using our script. Again, this helps retain star luminance & color.
2) This script was also designed for people who dont use Star Xterminator (SXT) i.e. Starnet++ and Starnet 2. Example, if someone did not buy SXT and used Starnett++, how do they properly remove the stars?? The substraction method used in Starnett++ is not the right way to do things unless the data is linear but Starnett++ works in non-linear mode so this is a problem. Also, even if you own SXT, our method of removing stars per(1) above is the better way of doing things, let me explaing more below.
3) The proper math to remove stars via "unscreen" mathmatics is (Starry-Starless)/(~Starless). The issue with this math is you can get errors when dividing by zeros. This becomes an issue if you have an image with a blown out core, like M42 the Orion nebula which was showing 1 values for the areas blown out (the inverse of 1 is 0), so you would get errors in the stars only image because Pixinsight would return the divide by zero error into a 1, so now your stars only image would be corrupted in areas which were blown out. SXT uses this formula for "unscreening" to avoid dividing by zeros (Starry - Starless)/(1.005 - Starless) which they create an offset to the data, but SXT also says to not use "unscreen" process for star removal on linear data, because a) the 0.005 offset they use in the formula to avoid dividing by "zero" could make for a big change or disortion when later processed; b) if you dont use the inverse of this formula to add the stars back, then you are working with apples & oranges and things may not match properly. So SXT says on linear data to just remove the stars with "Unscreen" option unchecked to avoid issues, which uses subtraction instead. In pixinsight, the easiest way to manage unscreening stars is to use this formula "rescale(Starry,Starless,1)" which will manage the divide by zero issue for us. So because this divide by zero issue is fixed, in our script we can benefit from this on both linear and non-linear data, thus we can control the exact methods of star removal the same way as when we add them back in as long as you use our script for both process.
I also want to make something clear to everyone who reads this comment. SXT is a very good process for removing stars from an image and is probably the best. It is because of this starless image SXT can create a stars only image by using the original starry image and using subtraction method for linear data (unscreen box unchecked) and unscreening math (unscreen option checked) depending on what option you pick per the data type, which is why Russel, whom I consider a friend, offers these options. But people may not know what option is best to use and they might be unscreening linear data when they should not be. The thing is, a lot of people also process in non-linear mode, because in non-linear mode we can do things like star reduction or narrowband normalization, Ligtness management, etc. which you cannot do in linear state. Times are changing and processing images in a non-linear state is becoming more common. So, when removing stars from a non linear image, we can still use SXT for the starless image only, but then use our script to properly handle the "Stars only" image part as our method of doing this is more controlled as described above, espcially when we screen the stars back to a new starless image. If you were to test this on an image with heavy nebulosity, you will see the difference.
So in your case, if you are only using SXT and the star only image it creates, in a linear state, and you process this in PhotoShop, this is perfectly ok as long as your "unscreen" check box is unchecked in SXT. However, we still think it is better to remove the stars on linear data using our script; Example. What if your "linear image" is over exposed i.e. say you imaged M42 at F2 with longer exposures and the nebulosity can be clearly seen not streetched. STX says to not unscreen if the data is linear, but in this case I would because of how bright the data is. But even if you did, you would get better results using or script because even though it is "linear", we have the option to reverse stretch it thus treat it as nonlinear, do the reverse stretch re (1) above, and give you better star output especially in areas of heavy nebulosity (as shown in the video). And for people who do not use SXT for the starless creation, or for people who work in non-linear mode, our script is what you would want to use to remove and add the stars. Its all about controlling the workflow and I guess I need to produce another video to detail this better. I hope all of this makes sense now. :-)
Please read this write up that Mike Cransfield did on subject on the Pixinsight forum to help explain things better:
pixinsight.com/forum/index.php?threads%2Fnew-script-screenstars.21098%2F
@@anotherastrochannel2173 many thanks for the comprehensive reply! And btw impressive work with the script. Quoting Mike's comment in the PI forum "For ScreenStars, therefore we use the unscreening formula for both linear and non-linear data. And, in keeping with Objective 1, adding stars back is always done by the conventional screening formula whether linear or not.", I conclude that for my workflow it would be best to use the scrip (instead of STX) to extract the stars from the linear image, then follow my normal routine of stretching the stars only image in PI and screen blend it as a top layer in PS. The "Reverse stretch" method is only used for processing already stretched images.
Wow, your answer il so clear ! Thanks@@anotherastrochannel2173
Hi guys
All of your tools are amazing and I hope that everyone is watching this video until the very end and do a donation to encourage and support you guys. You deserve it!
Many thanks
Bert
Thanks Bert! and for your support!
Great work guys! Nice to see your work is released now :-) Keep it up!
Thanks Luke! Should have a few more very soon
Your work is brilliant, Bill! I am eager to try out your star removal method because in AP we often see stars as something to get out of the way and overlook the beauty they have in their own right.
I agree, sometimes I like the star image better than the rest. The brighter the target is, the harder it is from a color stand point. The reverse screen helps with this but like everything else in life, nothings perfect, but it does help.
Again, I am suitably impressed with your wizardry. I was able to control bloated OSC stars that were taken with a UV/IR Cut filter of M42. One thing that I found was using Linear images, I was unable to select the "Enable reverse Stretch" using Version 1.1. Thanks to both you and Mike.
Fantastic work, many thanks Bill and Mike. I’ve just used this on the Elephant Trunk and North America Nebulae and it’s worked great to bring out the star colours and reduce the brighter ones to really show the Nebula detail. Clear skies Ian
Thanks Ian!!! Glad it helps you 🙂
Fantastic. This is the beginning of a great scripts family
I hope so, lol.
Already tried it. Love it. Ironically, I used my image of the Lagoon and Trifid to try it out.
Glad you liked it!! 😀
Thanks Bill Great work.Happy to buy you both a few coffees or beers what ever suits!!!!
That was very awesome of you!!! Thanks!
Great stuff Bill! Can't wait for you to work your PM magic in Siril! 😉
Thanks, we are looking into that too 😎
Awesome - I love that this is a script! ...and brilliant. Thank you for your continued contributions to astro photography processing - also, great image of the Lagoon and Triffid
Thanks Ed!!!
Thank you for that cool tool, what a very nice new approach to do that. I am curious what will come next🌟
Thank you!
I'm having fun with your script and I see a difference in the saturation of the star colors with your ScreenStars script.
I compared it with your PixelMath expression " RGBstars to NB V11 ".
I think I'll remove this PixelMath expression from my list of icons set on my narrow band workspace.
Thanks for the script Bill
I’ll support you from your donation link for sure.
Christian, actually, "for" true color replacement, the RGB stars to NB pixelmath script is the best way. In fact I recently updated the math (V14) but this time I am converting everything to a CIELAB color space because the lightness control is better then using luminosity. We plan on making a script for this too. As for the Screen Stars script, it is a better way to transfer stars over, but if the new target brightness is to bright or too extreme of a color change, you may see some discoloration still, but its better than using the conventional screening method. Hope this makes sense.
I see that this narrative is an overview of v.1.0. My copy, v.1.1, no longer includes reverse stretch option. Nevertheless, this script is an easy and painless way to achieve effective star reduction. Thanks so much, guys!.
fantastic work, thanks Mike and Bill 👍👍👍
Thanks!
Count me in brother! Amazing!
Thanks man!!!
Really wonderful Bill !!!
Thanks!
Great tool! I was wondering if you could add an enhancement to it to remove small stars. I know there are tools that reduce star size, but then you loose a lot of good looking stars. I have images where I want to keep the stars but not the star dust if you will. Very nice script!
Looks incredible, thank you 🙏
Thanks!
This is amazing, thanks so much for the script and the video.
just to clarify, the image with stars you loaded at the beginning of the process - was all stretching and editing done to the nebula with the stars in? should i remove the stars at the beginning then carry out the stretching/curves etc or do i do the whole process with the stars in the image, then do the screen script?
thanks again for your help, love your channel
Bill, I'm monitoring this for a response too....
Great script, I just tried it and it works well
Thanks!!
Awesome, Thanks, Pixelmath makes my head spin 7 ways to sunday.
Thanks!!
Excellent 👌…super boulot..! Bravo.merci à toi encore.
Thanks!!!
Hi Bill, You are brilliant. I appreciate your work. Thank very much. Joe D
Thanks Joe!
Thanks guys.. I'm gonna give it a go 🙂
Thank you for the script. I am exited to use it in my process. Do you use RC Astros StarXterminator or StarNet++ to create your starless image?
I actually use either SXT or Starnet2. Both do a great job but it all depends on the target. The best part of SXT is that it is currently supported and continues to improve, whereas StarNet2, I'm unsure of. Both give bot a try. Regular Starnet++ leaves artifacts so I don't use that at all anymore.
@@anotherastrochannel2173 Thanks. I am curious about the difference between the UnScreening of stars that takes place in STX vs the StarRemoval of your script? Thank you so much for all you have worked through.
Even when you are not using the reverse stretch option - ie just "conventional" unscreening - there is a small difference between the ScreenStars script and the RC-Astro SXT unscreened stars. In SXT there is a small adjustment introduced designed to compensate, in a rather approximate way, if your starless image has significant bright clipped pixels (i.e. blown out nebulosity). It does mean, for example, that if you combine your starless image with the SXT unscreened stars using the familiar screening formula, you won't gat back to where you started! ScreenStars uses a "true" unscreen formula, i.e. the precise inverse of screen blending. If your data is in reasonable shape, using ScreenStars is definitely preferable. If it's not in reasonable shape, it's best to go back and get it in reasonable shape and then use ScreenStars!
Awesome work!
Thanks!
I'm assuming you're using StarNet or Star Xterminator to create your initial starless images. Each of those processes creates (or has the option to create) the "stars" image. It would be nice to show the differences in what your script does when creating the "stars" image vs what those processes do by default. I think you kind of hint at it in the beginning of this video when you show how the "reverse stretch" option works. This is definitely a neat little tool, and I'm going to grab it to play with it... see how it compares to my typical processing workflows. Not like I'm able to actually image anything with the absurd weather we've been having... so might as well play with some old data LOL!
Jonny, so Starnet subtracts the stars, this is bad unless you are working in linear. SXT using a regular unscreen method. Our method we are performing a reverse stretch of the non-linear data, then unscreening, the returning back to non-linear which preserves the star detail in areas of heavy nebulousity. The comparison I did in the video between regular unscreening and also the reverse stretch method will be the same as our method vs SXT starless image. Play around with it for sure, and I completely understand about the weather situation. We finally got a break in the past 2 weeks for decent weather, before that it has been horrible. Its actually clear tonight, but too tired to do anyhting, lol.
Dear Bill
I am sorry but I don't understand the basic thing: Why do I need a starless image (RGB_starless) when I want a "star only" image from that RGB image? The RGB_starless image was made by another star removal process like SXT. So when I have only a starry RGB image and want to get a star only image from that, why do I have to select a starless image from the starry image? Sorry again for not getting the point! :) So your script is not an alternative to SXT resulting in a starless and a stars only image? But looking forward to the other scripts and appreciate your ongoing effort in releasing awesome scripts! Thanks a lot, Nico
Because starry image minus starless image equals stars only image. but instead of "minus" we unscreen the starry image from the starless to get the stars only, but to improve this even further, we perform a reverse stretch on the two data sets to a pseudo linear state, then unscreen, then stretch back, to give us a stars only image. This method produces a better stars only image. When we do this same reverse stretch screening process to transfer the stars over onto a new starless image, we get better star color protection. Re SXT, SXT is only unscreening the stars from the starry and starless image to give you stars only image (behind the scenes), we are doing it better by offering a reverse stretch. Please compare to see the difference. I actually showed the difference in this video from conventional unscreening and performing a reverse stretch unscreening, you will see the same thing. Again, in the video I expressed the big "difference" in both methods will be seen in areas of heavy nebulousity.
Here is the data I used in the video if you want to compare as I did :-) : drive.google.com/file/d/1Tz-XZ5jlI1s9TEhTx_f6FSfgHNwNeRvU/view?usp=drive_link
Dear Bill
Thanks for your clear explanation and the data! Of course, now I got it!🙈☺️ And I will include it in my workflow - but will still use - as I read - your other script in case of inserting rgb stars in a nb image.
One last thing: I often struggle stretching a linear stars only image just to the right amount regarding star size, brightness and halo radius/intensity so that the stars match the separately starless image afterwards. I wonder if there might be a pixel math stretch process in future... ☺️🙏
Great tool but I'm confused, for your Star removal you've already a starless image, so you've already removed the stars using..?
1) We are removing the stars in a better mathimatical way. 2) There are limitations to using Star Xterminator (SXT) when it produces a star only image; example: sometimes SXT leaves part of a star on the starless image which has to be cleaned up, the problem becomes, how do you get part of that star back onto the star only image?? The answer is to not use he SXT star only image, clean up the staless image by removing the star remant, then use our script to properly produce your stars only image. Does this make sense now?
@@anotherastrochannel2173 I think so, so in steps, 1. Create a clone of the image 2. SXT the clone, dump the star only. 3. Clean up the Starless image, 4. Use your script to produce new stars only image?
@@andythilo. Yes sir!! 😎
@@anotherastrochannel2173 Thanks I'll give it a go!
Again great work, works perfect!. Thanks a lot that you are shraring your work with us.
BTW. The link for Color Mask PixelMath dosn't work
Thats odd, it opens on my end: drive.google.com/drive/folders/1HOU2zqRfqArDtKPQiq67_q-JcZWnq2lK?usp=sharing
I've read in a few places that you might remove stars while in linear state, then process background and stars separatedly, and close to the end place the stars back. What I could not find is any tutorial/advise/workflow for processing the isolated stars. Anyone can help?
Great stuff !
Thanks!!
Another one 🔥🔥
Thanks!
Why do i see "view geometry differs" on the preview screen of the script instead of an actual preview? Also the script fails to run. I star aligned the "stars" image and the "nebula" image.
Hello Bill, I wanted to know how to remove the stars using the star screen script in the first place because when I opened the scripted It is a requirement to have a starless image already. Am I missing something or should I use starXterminator to remove the stars?
Thx for the great tools you are providing..
Yes, use SXT or Starnet2 to remove your stars to give you a starless image. You dont need to use the SXT "remove stars" feature, you will do this in our script as the way we do it using the "reverse stretch" math is a better way. Watch your new Star Reduction Script video which I just released to see how they both work together.
@@anotherastrochannel2173 thanks Bill 🙏will definitely check it out.
hello
I am using Screen Stars Script on Mac and Enable Reverse stretch box is grey. can not tick it . please help what I am doing wrong. many thanks Krzy
Are you working on linear data? If the script detects that the image has not been stretched, or only very lightly stretched, it will not try to do a reverse stretch and the option will be greyed out.
Is it possible to modify this script to allow rescreening of mono stars onto a color image? Sometimes mono stars are used for artistic effect but the tool in its current configuration doesn't like that combo.
I took a stretched OSC image, did a star removal, and then star replacement (both with reverse stretch enabled). And I created a star image using Starnet2 and did a simple starless+stars Pixelmath. And both results are identical. So am I missing anything or whats the advantage or usecase of the Script?
In the video I said that both images will look the same in areas of no nebulosity, but in areas of heavy nebulosity, you will see the difference, as I shown in this video. Test this on some data with heavy nebulosity. If you dont have any data, I can send you a link to test on 😀
Here is a link to the data I used in the video: drive.google.com/file/d/1Tz-XZ5jlI1s9TEhTx_f6FSfgHNwNeRvU/view?usp=drive_link
I'm have a problem loading your screen stars script. I've added the url several times. PixInsight isn't picking it up with updates
Is this the URL you used? also, make sure you click "Update" to install and reboot Pixinsight
www.cosmicphotons.com/pi-scripts/screenstars/
I am seeing the same thing happen. Added the repository, checked for updates, installs the script, restart PixInsight, but no "Screen Stars" appears under Utilities
Hi Bill, Thank you for your continuing help to the astrophotography community. I have followed your instructions to install this script and in the Installed Updates page of Manage Repositories it shows up but I cannot find it in the Utilities drop down..... I notice that in the installed list the script shows as a single line with a scroll type icon whereas with the GHS script for example it also has other lines one with a gear icon. I wonder am I missing something? I have tried to repeat the installation with no luck. Any help to a PI newbie would be most welcome. Thanks again
Did you use this address? www.cosmicphotons.com/pi-scripts/screenstars/
In the installed updates page the scroll icon signifies a script, the gear icon signifies a process. GHS exists as both a script and a process, hence multiple lines (also a documentation line). For Star reduction you should have a scroll and a document line. I don't know why it is not appearing in Script>Utilities menu. Hopefully the files are there and you can install manually via the Feature Scripts utility - assuming you know how to do this?
Thank you for the replies gentlemen. I have successfully installed the Star reduction script with no problems at all. In the Installed updates list for Screen Stars there is only the one line, no documentation line as with Star reduction, is this correct please? The address I used was copied as your demonstration Bill and when I look in the Feature Scripts list Screen Stars is not present Mike. Sorry to be a pain, thank you.
@@petecarpenter1068 It is correct that there is no documentation line for Screen Stars - we did not feel additional documentation beyond the comprehensive tooltips was necessary. I don't know why Screen Stars is not in the Feature Scripts list - one option would be to find the updates.xri file in your PixInsight directory and rename this (eg as updates.xri.old) then restart PixInsight - it should then reload all of your PI updates including the official PI ones as well as Star Reduction and Screen Stars.
A aligment is still nececarry?
Yes, data must match.
How the heck do you enhance the dust in the background. Even with a lot of hours, my background just shows noise, and I can never get dust, just removes the noise and I can't figure it Out.
I'm not sure what your setup is, but this was about 4 hours shot from bortle 8 on a 80% moon lite sky using a Baader Moon and Sky Glow filter, Edge11 with F1.9 Hyperstar using the PlayerOne Poseidon OSC camera. Make sure you shoot broadband in order to see the dust 😀
@@anotherastrochannel2173 ok, then I need to learn how to combine narrowband OSC data with broadband to pull that off. Then maybe range selection to seperate the narrowband object and blend it to broadband background. . Thanks!
Sorry but I'm even more confused than I was before watching the video. Why do I need a starry image and an image of just stars to remove them from the starry image. I must have done that already ?
You only need a starry image and a starless image to remove the stars. To add stars to another image, you will need the stars only image and the new starless image. You might want to watch the video again and then you will see what we are doing 😀
Just wondering why the url for screen stars keeps timing out...I would love to use that script for my images.
There does not appear to be any issue with the repository now - may be just a transient issue when you tried. Have you now managed to get the script?
Génial 👍….!
Thanks!!
at 04:35 "I already created a starless image". How? With this script, some other script? I want to remove stars from an image with foreground galaxy and background stars. I'm looking at the script GUI and I see in the files section: Starry View, Starless View and Stars view. Can I, and if can, how exactly can I load my xsif file of M106 and background stars in it, remove the stars from it, generate a starless view of the galaxy and a "Stars View" which I assume is only stars? Sorry, I really appreciate your work, but the whole reason I installed this script in PI was to use it. not starnet2 or other star removal plugin, to remove stars from existing image and generate the equivalent of a star mask and a starless image. That information is absent in this tutorial.
Will this work with stars extracted using StarExterminator?
I dont see the point in using SXT stars when our method of star removal is a better method. However, if you dont want to use our star removal method and just use SXT stars to add to a starless image, you can use our ScreenStars script "Star Replacement" option and dont use the reverse stretch option, as this uses is simple screening math.
I watched the video again, and realized that this tool might be only for OSC (one shot color) camera data. I shoot mono with filters, so would this tool be of any use for me? If so, then what are the workflow steps for mono data?
It is good for RGB data, either OSC or Mono combination
Hi, thank you for all the work you've done. I've read through all the comments here, and I, too, am totally confused. So your script needs both a starless image and the original with stars. How exactly are you supposed to get a starless image without first removing the stars from the original image? Isn't that what the title and description of your tool is, "This script provides a tool for extracting and recombining stars." ? Also, what is meant exactly by "screening" and "unscreening?" I see those terms thrown around a lot on various videos with no explanation.
If you have to use another tool to extract the stars from a complete image, then what is the point of your tool? Please explain to a total dummy. (I think there are a WHOLE LOT of people out there like me earnestly trying to learn the deeper levels of Pixinsight). BTW, I enjoy learning Pixinsight, but sometimes the frustration gets the better of me. I do SHO and RGB with a monochrome astro camera, with mainly cookbook type workflow gathered from TH-cam sources like yours, constantly evolving as I learn more. Thanks!
You create the starless image using StarNet2 or Star Xterminator, then you run our script. Once you have the starless image, you can use our Script to properly remove the stars. You do not want to use Starnet2 or SXT for removing stars as our script does a better job of it. So after you use our script for removing the stars, you can later use our sctipt to add the stars back to your modified starless image.
I guess I need to do a better job of making videos. I will do a follow up video on this next week. I do suggest you watch some of my other videos, like adding RGB stars to Narrowband, etc, as I do cover some of the math re adding/substracting vs unscreening and screening, this might help you understand.
@@anotherastrochannel2173 As an engineer with 25+ years background in aerospace and backend database software, I can certainly understand the old adage, "the more you know, the more you realize how much you *don't* know." I always had a vague knowledge of image processing, using basic tools to edit regular photos, like contrast, brightness, hue, etc., but never really knew how any of that worked at a software coding level. Coming into Pixinsight with that little bit of knowledge about a year ago (also minimal Photoshop experience at the most rudimentary level), it seems to be very much geared to those with an expert level of knowledge of image processing. I think there are multitudes of people out there like me, who wanted initially to apply a simple cookbook recipe approach to processing astro photos (I jumped straight into narrowband/monochrome without going through OSC first), with workflows gleened from Pixinsight TH-camrs like yourself. I can do pretty well and make impressive final images, but I don't really know why or how I am doing it--I'm just following others' recipes and tweaking the processes for my own data for the most part.
At some point, I wanted to gain a much deeper insight into what I was actually doing with all these tools. That's where my issue arises. Pretty much all these TH-cam tutorials assume you are either a total newbie with no knowledge of image processing (and they don't give you any detailed explanation), or you are a total expert (and they don't give you any detailed explanation). There seems to be no material geared toward the intermediate level user. It can be very frustrating to watch tens of hours of TH-cam videos just to get that 30 second nugget of knowledge that leads to that "Ah ha!" moment when it all comes together. I think I speak for many tens of thousands of astrophotographers when I say that.
Concerning your "star removal tool," my take on it is that you are taking advantage of another tool already in common use that produces a starless image, which you then backwards engineer somehow with the original image to get a "better" stars-only image. I think the main question/confusion that people are having (like me) is, why don't you just make your tool so that it extracts the stars in the manner you want from the original image, without needing yet another tool to create a starless image? This explanation would be extremely beneficial to an engineering-minded person, which is what Pixinsight seems to be all about. After experimenting with Starnet and SXT, it's obvious that those tools don't just "remove stars," since it would leave darks spots on the starless image where the stars used to be. It actually fills in those holes with some sort of AI voodoo magic process--not perfectly though, because on my data, it sometimes leaves little (and not so little) puffs of brightness from the halos, which I then have to use clonestamp to brush out. This has led me to much experimentation on *when* and *how* to apply certain tools in the workflow, and it all makes a huge difference in the quality of the starless image. A huge difference.
The reasons for the order in which things need to be done are never really explained well in most videos, and I have wasted many days trying to figure which process order gives the best results.
Another issue/question I have with star removal is the inability to run any kind of SPCC or PCC on the linear, combined colored (SHO or RGB) starless data in order to correct the colors. These tools cannot run without stars in the image, as far as I can tell. So, what is the point of SPCC if you can't run it on a starless image? It seems like an impossible situation. And it only works on linear data.
These are the sorts of things I run into almost every time I try to learn and add a new process or script into my workflow, resulting in a complete upset of the order in which I do everything. I might have 15 or 20 different steps, depending on the object, quality of data, etc. It often seems as if there is no rhyme or reason to what some Pixinsight TH-camrs are doing compared to other Pixinsight TH-camrs. For someone who is at an intermediate level, and unable to really explain what I am doing at a fundamental level, it can all be extremely confusing and confounding. I simply am not able to make any judgement as to which approach is correct, other than many, many hours of trial and error. That is not very fun.
Another killer is not knowing which processes work only with linear data, and which processes only work with stretched data. That always gets me, and I have to constantly return to the drawing board with my entire workflow, reinventing the wheel over and over.
The most helpful video would be to watch an "expert" take some image data that he had never seen before, and do a processing video without rehearsing anything. Watch what he does when problems arise. That's where 90% of my learning comes from.
We need some sort of apprenticeship program for Pixinsight. It's the most unintuitive pursuit I have ever attempted in my life (other than learning UNIX and getting along with women). Thanks for all your work, and I certainly understand that people are taking a risk putting themselves out there on TH-cam as a useful service and get all sorts of guff. But I just wanted to reflect what at least tens of thousand of people are thinking about Pixinsight and learning its various tools and methods. Cheers!
Hey Bill. I am using a mac and when I open the script Iand enter the two views, I don't get anything in the preview window. Any suggestions on what I am doing wrong? Thanks
Are using using a Starry image and a starless image that was directly made from that starry image?
You lost me, how did you remove the stars in the original image? I have a starry NB image and I need to remove the stars from it and later replace them with RGB stars. How do I do that? When I put that image into the script and select "star removal", and hit the check, it just fails. What did I miss?
Its all in the video. To remove stars from an RGB image, select star removal option, load your RGB starry and RGB starless image, then remove the stars. use the reverse stretch option too if your image is non-linear; this will create your "Stars Only" image. To add the RGB only stars image to your new narrowband image, select the star replacement option, load your stars only and narrowband starless image, use reverse stretch if data is non-linear, then create the image. Note: Make sure your images have been registered at the beginning of the work flow so they match image size and have propery star placement.
@@anotherastrochannel2173No, actually it's not. This confused me as well, but I think I know what you did. What he is asking, and what you didn't show, is how you got an RGB starless image to use with this process. I believe you used SXT to generate the starless image and you toss the Stars image that SXT produces. You then use the starless image that SXT produces. I'm realizing that this processes isn't a replacement for SXT or StarNet for the purposes generating a starless image, rather it's a replacement for those two processes for the purpose of producing better stars to add back into your starless image. It's an awesome process, and absolutely amazing contribution to the community, but it could use better explaining, like from a workflow perspective. I envision using this immediately after my initial stretch after (graXpert, SPCC, BlurX) and at the very end when I need to add stars back in, followed by star reduction.
I tried to use your script.I have two non linear images of the same subject, one with 2 hours exposure and the other one with 16 hours of exposure time. From the first one i want to use the stars and put them on the 16hours starless image. I got the star image in reverse stretch mode and it looks good, but it does not let me replace the stars with the script. apparently i'm doing something wrong. Maybe somebody can tell me what i am missing?
Did you register both images together so they match in size? If the images are not the same size then it won't work so you need to make sure that you register one image to the other
@@billblanshan3021 Thank you so much for your quick answer. i did the star alignment, and it worked. 😊
Hi there. I am a bit confused. At the Linear example you adding stars to a linear image which has been altered with NBNormalization. I thought that NBNorm should be used on stretched images. What am I missig. I am using StarXTerm at the linear stage and putting them back after stretching, but I have troubles to stretch the "OnlyStars" image properly without blowing them out. STF+HistTransf is very tricky for onlystars. What tool do you use for stretching onlystars images? Thanks a lot
Tom, You cannot use STF to histogram transformation on a stars only image because the STF is stretching an image to a known median background value thus it blows the stars out. So if you're going to use STF then you need to reduce the mid-tones bar so the stretch isn't so extreme, otherwise don't use STF and just use histogram transformation to stretch your stars.
I personally do not do any processing in linear stage because of issues like this. There are many advantages to processing a non-linear using narrowband normalization but one of the reasons why we did the screen stars script is so that you can remove the stars from a nonlinear image and use the reverse stretch option to remove and add stars back to an image this have all the advantages of removing and adding stars from a pseudo linear stage perspective but doing it on a nonlinear image.
Much Thanks Bill for replaying to me, you are doing hell o a job, crating new things and explaining them both.. I am still learning, RC-Astro recomends to remove the stars as soon as possible (before stretching) so I was bit confused.@@billblanshan3021
BTW do you guys know how to reproduce EZ soft stretch in pixinsight? It was really easy to use, now its gone... sad,sad,sad...@@billblanshan3021
If you’re doing it non-linear why is there a need to remove and re-add them if you stretched the image with them? It looks like you have 2 stretched images and you just removed and then added them back? Please tell me what I’m clearly missing haha.
I understand doing it when it’s linear because when you remove the stars, you can get a better stretch on the nebula without having the stars getting blown out and what not but if you’ve already stretched the data, why are you removing them and adding back in?
Clearly, there’s an advantage I’m just missing what it is.
Is it strictly for putting RGB stars on a non-RGB image? In the first example it look like to RGB images so that’s why I didn’t make sense, but then I saw later are you adding RGB stars to an SHO image?
Yep, removing your original image and then replacing them on a different image. That's the goal but doing it in a better way
@@billblanshan3021 would you recommend removing them before stretching in certain cases to get a better stretch or do you always stretch them with the stars still there?
@@rvoykin You really have a choice, you work in linear phase or in non-linar phase. I would not mix. Wait for my next video, you will get the idea of where we are going with this 🙂
Thank you so much for the script!
You are very welcome!