I have struggled quite a bit with getting both deep imaging of nebulousity or galaxy arms and at the same time maintaining good star cores. I would invariably get ugly artifacts in my stars and multicolor halos. Its been very frustrating. I have tried RGB star replacement with varying degrees of success. I have also started doing HDR image capture so that I can eliminate star core saturation and still be able to get better depth in the nebulousity. But it was very hard to get good non-linear stretching without creating other types of color artifacts in the stars. However, I have been trying to test a new technique of combining HDR image capture with the use of the Repaired HSV Separation script in PI. The first time I tried this it was amazing. I went back to my data set for M13 as a test of the script. This was not an HDR image set, but I have not been able to really complete a full target yet with HDR for all channels since the weather has been horrible for the last two months. I really liked the M13 data and I thought it was be a good test. It was 12 hours of integration on a bright target. My processing was not bad, but the star artifacts really distracted from the image since stars are what you are after in globular cluster imaging. The original processing and all the details are here:
What I did was the following test. I went back to the project and opened the process history for the final image and created a process container from the history. I deleted all the steps prior to stretching the image to non-linear so I could apply the exact steps as the original image to the test images. I then went back into the final image history and reset the image to the step just prior to the first non-linear stretch. Since the Repaired HSV Separation script will output new mono images of the separated channels and not alter the original, I could apply it to the original image and not alter anything. I ran the script three times with the repair levels of 0.25, 0.50 (default) and 0.75 to compare the effects of the script. I then used the process container that I created to re-apply the exact same subsequent processing to the new HSV repaired image as the original image so that I could compare apples-to-apples. I found that the default setting for the repair level in the script worked the best with the 0.25 level under corrected and 0.75 level over corrected/washed out. But the stars in all three where MUCH better than the original image.
This is the original image:
You really need to click on the images to see the stars close up, but the improvement in the stars is quite stunning. By using the process container, the ONLY difference between the images is the effect of the Repaired HSV Separation script. I am going to go back and reprocess some of my previous data sets where I really could not get the star fixed. I think that this is a big breakthrough.
Comments and Questions are welcome.
Edited by cfosterstars, 16 March 2019 - 09:34 PM.