M51 in HaLRGB-Pixinsight

Processing M51 in HaLRGB

The data for this project was acquired in my backyard in April of 2024 and 2025, with my 6" RC scope, ZWO 294MM Pro, and ZWO 1600MM Pro cameras. The data break down like this:

2024 (294MM Pro):

110x75s Luminance, gain 50, offset 15, -15C
35x225s Red, Blue, and Green, settings as above
55x225s Halpha, gain 121, offset 30, -15C

2025 (1600MM Pro):

780x30s Luminance, gain 100, offset 30, -10C

I combined the lum data from both years, but I used the wrong RGB filters in 2025. Oops.

All of the data was pre-processed and stacked into master files using Pixinsight.

Read about combining the 2024 and 2025 luminance master files here.

Linear processing and stretching was done in Pixinsight and the resulting masters where combined in Photohop.

Master Files

The master files were aligned together and cropped. The flats for the 2024 data did a poor job of correcting the gradients, as you can see below.

Red

Notice the dark center of the frame surrounded by lighter edges, due to poor flats correction.

Green

There is also an uncorrected dust shadow in the upper left.

Blue

This closeup of the blue master shows more circular dust shadows in the lower right background.

Cropping

I decided to crop this image tightly to reduce the impact of the poor flat field correction, namely the uneven background brightness and the dust shadows.

Combined Lum

After alignment and cropping.

Ha

This is the Ha data after PI processing.

Gradient Removal in Pixinsight

The first step is to remove the complex gradients from in the master files. Cropping helps, but there is still more to do. Graxpert makes quick work of the gradients.

Graxpert Settings

Lum after Graxpert. I use division on mono images.

RGB before Graxpert. The green color cast is caused by light pollution.

RGB after, with GXT set to subtraction instead of division.

Color Calibration

Next, we'll use the Color Calibration and Background Neutralization processes to ensure the accurate color. CC requires two previews- a small one to capture the sky background and a larger one to provide a white reference. Galaxies are made up of the combined white light of billions of stars and make excellent reference objects.

We can use the same background preview for the neutralization process.

Before color calibration.

After calibration and Background Neutralization

Deconvolution (Sharpening)

Deconvolution analyzes the shapes of the stars, which "should" appear as single points, to create a model of blurring caused by atmospheric turbulence. That model is then applied to the image, restoring some detail. Blur Xterminator is the tool of choice.

Lum before BT.

Lum after. Wow!

Settings.

Stretching Lum and RGB

Now it's time to permanently stretch the images. I'll simply adjust the screen stretch, if necessary, by dragging the STF sliders, then apply it using Histogram Transformation. Here, I moved the blackpoint and midtone sliders to slightly darken the background of the RGB image.

Lum Noise Reduction

The next step is noise reduction with Noise Xterminator. Its most recent update, V3, is simply amazing. NX lets us dial in noise correction with different adjustments for high frequency, low frequency, and pixel scale. High frequency noise occurs at the level of just a few pixels, while low frequency noise appears as larger blotches. We can see each type of noise by setting its slider to 0 and maxing out the other.

This shows all of the noise.

By moving the HF slider to 100 and the LF to 0, we can clearly see the low frequency noise in our image.

Conversely, let’s move the LF correction to 100 and HF to 0 Now we can easily see the HF noise.

Now I adjust the sliders until each type of noise disappears. I try to keep the LF reduction as low as possible to maintain contrast.

We can find the correct scale value by increasing HF correction to 100, setting LF to 0, then moving the scale slider until the noise we see is about the same size as the details in our image.

The noise reduction effect is supposed to improve with more iterations, but I find that higher values sometimes create background details that aren't really there.

RGB Noise Reduction

Noise can often by worse in the RGB image due to the lower integration times of the R, G, and B masters and the addition of a color noise component. We can use a similar procedure as above to clearly see the color noise.

RGB image with no correction.

Large scale color noise.

Small scale color noise.

RGB with corrections.

Coming Soon: Taming The Galaxy Core With Multiscale HDR

The galaxy core is very bright in both images, which will cause us to lose color and detail when ccombined. Multiscale HDR can correct this. More information about it soon.

Star Removal

Separating the stars from the image allows us to apply processing steps that enhance the object, in this case a galaxy, without hurting the appearance of the stars. There are several tools to accomplish this. Once again, I will go to the XTerminator toolbox with Star XT.

We won't save the stars from the Lum master, so we don't need any special settings.

We will want to save the RGB stars to add back in later, so we'll check Generate Star Image. Removing stars using an "unscreen process" lets us add them back in Photoshop using screen blending mode. Large overlap can improve the result but takes longer to process.

Preventing Too Much Xterminator

Occasionally, Star XT will get confused and mistake object detail for stars, removing it. This happened to our RGB image. To prevent that, we'll create a mask to hide certain areas of the RGB image from Star XT. I undid the first removal, used the Halpha image to create a mask, and then applied it to the RGB image.

Halpha image. I used the Halpha regions in this master as the basis for my mask.

The mask. The black areas will be protected from Star XT.

The mask applied to the RGB image.

Finally, We'll Save Our Master Files

It's time to move out of Pixinsight and into Photoshop. We'll save all of our images as 16-bit TIFF files. We'll combine five images in PS:

Lum Master

RGB Master

The "unscreened" stars saved as a TIFF.

We'll add this Halpha data to our image to make those regions more visible in our final image.

We won't add all of the Halpha data. Some of it isn't from Ha regions, but from other areas that were bright enough to shine through the filter.

We can subtract an image of the red channel from the Halpha image to remove some of those regions. This is known as "continuum subtraction."

We'll apply the Channel Extraction process to the RGB image with only R checked.

Red channelextracted from RGB for use with continuum subtraction.

Next we'll load our masters into a stack in Photoshop and begin processing.

Google Sites

Report abuse