Stretching the Color Data:
It is now time to stretch the RGB data using HistogramTransformation. You can delete the previews used in ColorCalibration.
One thing I want to point out before I forget is the lack of dust donuts here in the RGB image. The reason is that flats were taken for each color separately. If you take just an L flat, you will often be dealing with them here.
Again, we will use a HistogramTransformation, I have moved the middle slider quite aggressively. I am going to assume you know the drill by now of using the real-time preview and then applying the transformation to the actual image.
It is now time to stretch the RGB data using HistogramTransformation. You can delete the previews used in ColorCalibration.
One thing I want to point out before I forget is the lack of dust donuts here in the RGB image. The reason is that flats were taken for each color separately. If you take just an L flat, you will often be dealing with them here.
Again, we will use a HistogramTransformation, I have moved the middle slider quite aggressively. I am going to assume you know the drill by now of using the real-time preview and then applying the transformation to the actual image.
The result...
By the way, might notice that this result is more washed out than the STF we did earlier. The reason for this is that the STF when used in auto mode will happily clip the blacks.
By the way, might notice that this result is more washed out than the STF we did earlier. The reason for this is that the STF when used in auto mode will happily clip the blacks.
Now for TGVDenoise on the RGB image. Create some previews around an area with some background but some interesting detail ([alt]n). Also create a preview around the center of the galaxy.
With an RGB image, we can do denoise for both the luminance (lightness) and the color (chrominance) portions of the image. Again, the key to good TGV denoise is lower the strength way down and up the iterations to something like 500.
If I recall correctly from the excellent (and free) video tutorial by Warren on TGVDenoise:
www.ip4ap.com/IP4AP_Subscriber/Pixinsight_2/PI-13_Nonlinear-7/PI-13_Nonlinear-7_player.html
You should use RGB on linear images. On non-linear images, you should use CIE Lab.
Here are the settings I ended up with:
If I recall correctly from the excellent (and free) video tutorial by Warren on TGVDenoise:
www.ip4ap.com/IP4AP_Subscriber/Pixinsight_2/PI-13_Nonlinear-7/PI-13_Nonlinear-7_player.html
You should use RGB on linear images. On non-linear images, you should use CIE Lab.
Here are the settings I ended up with:
Make sure you have the local support checked.
Once applied, the image looks like this. You will notice it pretty much looks the same. And that is the point. At a macro level it should look the same. It is only when looking closely that the differences should become real apparent.
Now delete the previews. We don't need them. It is time for another histogram stretch.
NOTE: DO NOT use the auto zero shadows. It will do each channel individually and mess up the color calibration.
When moving the left slider, go as far to the right as you can without clipping the image (the pixel count for shadows should stay at zero).
NOTE: DO NOT use the auto zero shadows. It will do each channel individually and mess up the color calibration.
When moving the left slider, go as far to the right as you can without clipping the image (the pixel count for shadows should stay at zero).
Things are starting to look a bit better.
