Does anyone know of any video tutorials with masked stretching techniques for star only images? Once stars have been pulled out of an image, I am looking for techniques on how to stretch them and keep the brighter ones tight and not bloated while still bringing out the fainter stars. I do have Bill Blanshan's star reduction scripts for reducing stars at the end of the processing process but I am curious to see if there is a way to control my stretching a little more through the selective use of star masks.
Looking for guidance on controlling star bloat when stretching a stars only image after using StarXterminator.
Well Written Engaging
GHS with a positive "b" value should do it (for example). You could go all the way to 15 but I find that can be a bit extreme, especially when you start adding saturation. I typically shoot for around 4-5 on b. Then just raise the stretch amount until you're happy. You may also want to use the Color mode to get a proportional increase in saturation. If that's too extreme, you can always just do the RGB stretch as described and then a second, basic Saturation stretch.
Helpful
Hi Chris, I use GHS in the same manor as Roman describes above. After I finish everything in the linear state, I create a copy and use it to stretch the stars in GHS.
I usually taking the B value between 4-6.5. I then stretch the image between between 50-80% of the first quarter of the histogram. I also utilize the SP near nebulosity but not within it and the real-time preview to find the right balance. After stretching, I use starX to create a stretched star image and add a small amount of saturation. I finally add the new star image to completed non-linear image using pixel math use simple add expression $T+stars or screened stars expression with combine (ABE_P, star, op_screen()).
Below is the example I have completed with starnet2 prior to using starX.
Here is a quick example of some RGB data from the Gecko Nebula. It was stack that I did immediately after capture
Left is the method described compared to base star stretch on the right
Here is a grouping of stars near IC 2169 in Monoceros from a recent poor seeing night where I was only able to use 43 of 100 3 min subs. The GHS stretched stars are on the left and standard stars on the right. I think it illustrates the polygon result and its improvement with GHS.
Hope that helps or at least makes sense.
Clear Skies!
- Brandon
I usually taking the B value between 4-6.5. I then stretch the image between between 50-80% of the first quarter of the histogram. I also utilize the SP near nebulosity but not within it and the real-time preview to find the right balance. After stretching, I use starX to create a stretched star image and add a small amount of saturation. I finally add the new star image to completed non-linear image using pixel math use simple add expression $T+stars or screened stars expression with combine (ABE_P, star, op_screen()).
Below is the example I have completed with starnet2 prior to using starX.
Here is a quick example of some RGB data from the Gecko Nebula. It was stack that I did immediately after capture
Left is the method described compared to base star stretch on the right
Here is a grouping of stars near IC 2169 in Monoceros from a recent poor seeing night where I was only able to use 43 of 100 3 min subs. The GHS stretched stars are on the left and standard stars on the right. I think it illustrates the polygon result and its improvement with GHS.
Hope that helps or at least makes sense.
Clear Skies!
- Brandon
Helpful Engaging Supportive
I tend to stretch my images to about 25 or 30% using histogram stretch (or wherever the stars look like I want them in the final image) and then extract them with StarXterminator. That means I have very little else to do with the stars whilst I go about stretching the starless image. Then just use a pixelmath re-combine.
Helpful
Thanks all, I have tried these methods but found that I still wasnt 100% happy with the outcome when processing stars that have been removed from an image via StarXterminator. I did try something today on an image of NGC6188 which I have been re-processing with a new workflow and seemed to find an approach that might have some promise. What I did was pull the stars out with StarXterminator, run two fairly light stretches on the stars using Histogram Transformation to get the bright stars to a point where I wanted them to be for the final image. Once I had that, I used Range Selection to get a mask for only those brightest stars. I applied that mask to the stars image, and did stretches using Histogram Transformation to get the smaller stars to a final state. I then rescreened the stars with the Pixelmath below. combine(starless,stars,op_screen()) The result that was able to get is https://astrob.in/o0rern/D/ Curious on what people think of the approach and the result.
Helpful Insightful Respectful Engaging
HI all,
Of course I am going to offer up a GHS solution..... but how you do it will depend on whether you want high-contrast stars (with a Gaussian shape), or low contrast shape (round with little/no halos).
Using GHS with a high b (maybe a couple of stretches with HP to control brightness), will give the former, as suggested above.
For the latter, check the invert box in GHS, apply an intermediate +ve b, move SP (right up to 1 initially), and apply stretch to form circular stars. To lower b towards 0 and adjust D to achieve your desired star shape. This will leave the star brightness nearly saturated (so they won't really hold colour saturation). To reduce star brightness, reduce SP from 1 to get the desired star brightness. With the controls you can somewhat independently control the star brightness, size, and colour retention.
I often use the latter technique to create a mask from my extracted NB star_mask luminance to past RGB stars into my starless NB image. Alternatively you can blend a starry version onto a starless version of luminance image with the pixelmath: Starryimage = Starlessimage + (1-Starlessimage)*star_mask and then LRGB combine with your starry colour version.
CS & I hope this makes sense...
Dave
Of course I am going to offer up a GHS solution..... but how you do it will depend on whether you want high-contrast stars (with a Gaussian shape), or low contrast shape (round with little/no halos).
Using GHS with a high b (maybe a couple of stretches with HP to control brightness), will give the former, as suggested above.
For the latter, check the invert box in GHS, apply an intermediate +ve b, move SP (right up to 1 initially), and apply stretch to form circular stars. To lower b towards 0 and adjust D to achieve your desired star shape. This will leave the star brightness nearly saturated (so they won't really hold colour saturation). To reduce star brightness, reduce SP from 1 to get the desired star brightness. With the controls you can somewhat independently control the star brightness, size, and colour retention.
I often use the latter technique to create a mask from my extracted NB star_mask luminance to past RGB stars into my starless NB image. Alternatively you can blend a starry version onto a starless version of luminance image with the pixelmath: Starryimage = Starlessimage + (1-Starlessimage)*star_mask and then LRGB combine with your starry colour version.
CS & I hope this makes sense...
Dave
Helpful
Andy Wray:
I tend to stretch my images to about 25 or 30% using histogram stretch (or wherever the stars look like I want them in the final image) and then extract them with StarXterminator. That means I have very little else to do with the stars whilst I go about stretching the starless image. Then just use a pixelmath re-combine.
This except I use ArcSin stretch on the stars only image and recombine in Affinity with Screen. If the image was shot SHO and I don't have RGB stars, I will combine a version HOO just to extract the stars and do a photometric colour correction on them. It makes a much nicer colour rendition of the stars. Not using PI so not sure if the ArcSin function is identical.