Bringing out the details part 2: finding hidden structures - HDR Multiscale Transform (Pixinsight)
This post picks up where the last one left off. I'm going start with my M51 (The Whirlpool Galaxy) luminance image and show you how find some hidden structures in the core using a process called HDRMultiscale Transform (HDRMT) in Pixinsight. I'll also demonstrate the technique on the core of M42 (The Orion Nebula).
If you aren't familiar with HDRMT, it is a process that can help tame "washed out" regions of your image. To be clear, this doesn't mean "clipped" regions. For this context, I'm referring to areas of an image that are so bright they obscure details. This often happens in the core of a galaxy because it is much more luminous than the spiral arms. The core of M42 (the Orion Nebula) is a great example of a nebula that has a washed out core.
I'll first apply this to the core of M51. Here's the image I ended up with from the last blog post after applying LHE:
Notice how the core is relatively brighter than the rest of the galaxy. My goal is to bring out some structures that are hidden within it.
First, I'll make a "custom" mask using the GAME script - which can be found under Scripts > Utilities > GAME. If it's not there, you'll need to install it - but I'm pretty sure it comes pre-installed with the latest update to Pixinsight. I'm going to create a circular gradient mask using the script:
Here's the mask:
I'll apply this mask to my M51 L image and create a closeup preview of the galaxy core. If I open the HDRMT, the default settings look like this:
When I use this process, I focus mainly on two parameters: "Number of Layers" and "Scaling function". The best way I can explain the "Number of Layers" is that the higher the number, the larger the structure it will bring out. Smaller numbers will bring out smaller structures. I find that it works best between 3 and 7 depending on the target. For the "Scaling Function", I use B3 Spline (5) for most applications, but often use one of the small scale settings if I want to bring out even smaller details.
If I apply the default settings to my masked preview, here is the result:
Comparing this to the original, I think I need to lower the Number of Layers. I'll try changing it to 5. Below is the result:
Better, but I still see some details trying to work their way out. I'll change the Scaling Function to "small scale 32". Below is the result:
Now I'm seeing some nice detail. This is about where I want it - but I'm seeing just a tiny bit of artifacting, and perhaps it has made the core a little bit too dark overall. There's a very easy way to fix this by altering the mask. I'm going to open the Curves Transformation process, select the "linear interpolation" button and bring my highlights down about 70%. Curves should look like this:
I'll apply this to the MASK only. This reduces the intensity of the mask, and thus the intensity of the process on the image it is masking. I'll apply the same HDRMT settings to the M51 image with this less intense mask. Here is the result:
This is more balanced to my eye. I could adjust the intensity of the mask again if I wanted, but I'm happy with this. For reference, here is the original vs the HDRMT:
Now, let's try this on a nebula with an overly-bright region. The Orion Nebula is probably the best example of how this process can be applied to bring out hidden details. Below is my cropped image of M42. It's 21, 1-minute exposures from a bortle 3 sky. I've only color calibrated and stretched it for this example:
You can clearly see the washed-out region of this nebula. Thankfully, my sub-exposures were relatively short - so the bright region is not blown out. The first thing I'll do to uncover the details behind this region is make a range mask that covers that specific area using the RangeSelection process. Here's my mask:
I've made it "smooth" so that the stars aren't affected. Now I'm going to apply this mask to my image and create a preview around the washed-out region. I'll open the HDRMT process, reset it, and experiment with the "Number of Layers" and "Scaling Function" settings until I'm satisfied. For this example, the default settings seem to be the best choice (shockingly). Here's the result:
Now compared to the original:
You can see, HDRMT has revealed new details and even some stars that were not previously visible.
HINT: after applying HDRMT on M42, you might try a quick iteration of LHE on the same masked area to add some contrast to the region:
Enjoy digging up some details!