NGC 2244: The Rosette Nebula (Multi-Night-Image including Processing)

07 February 2022

It has been a while since the last clear night and so there were some changes I wanted to test but had not yet the chance to do so. These tests include

  • Using the D7500 instead of the D3500 to hopefully avoid getting colored rings.
  • Using a cheap ND-foil to dim the output of my flatfield box and by doing so getting exposures of one to three seconds.
  • Testing the new All-Sky-Polar-Alignment of the ASIAIR.

All the above went (mostly) very well so I now have a decent data set to play with. I say "mostly" because I must have forgotten to switch back to using ISO 800 after focusing my telescope so that I ended up shooting all frames with ISO 3200 for the first night. While this is by no means optimal, thanks to the very good sensor in the D7500 this is not as bad as it may sound.

For all of those that wonder if there is a part missing of the Rosette Nebula: Yes there is. As I don't (yet) own an astro-modified camera I do not pick up all the nice red color. That will have to wait for some time in the future.

The remainder of this article will provide a brief list of the steps that I took to initially process the image. The main purpose of this list is for me to have a goto list that I can use whenever I want to process an image in PixInsight and have forgotten how to do that. The post is by no means meant to be a fully-fledged PixInsight tutorial and is only meant to be a memory aid for me.

Processing steps in PixInsight

Selecting Data

Remove obviously faulty data

Use the blink tool (PROCESS > ImageInspection > Blink) to exclude clearly bad data (bias, dark, flat, light).


Select only the best data

Use the subframe selection tool (PROCESS > ImageInspection > SubframeSelector) to exclude subotimal data.

SubframeSelector: Setting

  • Select Measure Subframes as Routine.
  • Input the correct scale (1.33 arcsec/pixel for Nikon D7500); Use if unsure.
  • Select the correct resolution (bit per pixel; 14bit for Nikon D7500).

Look at the data and graphs and select the Approval expression:

SubframeSelector: Selection and output

The metrics I chose were FWHM and Stars and I try to tune the expression in a way that I keep around the best 75% of my data. Select an output folder where the selected frames should be placed.

Stacking the Data

After the bad data is excluded the stacking may commence.

Arrange the files

Arrange the selected lights and the corresponding files in a folder structure that supports the next steps:

- Night_1
  - Bias
  - Dark
  - Flat
  - Light
- Night_2
  - Bias
  - Dark
  - Flat
  - Light  

With the data arranged like this PixInsight is able to load all into the right sections when using the Weighted Batchpreprocessing script.

Run Weighted Batch Preprocessing (WBPP)

Run SCRIPT > Batch Processing > WeightedBatchPreprocessing. Input NIGHT as Grouping Keyword and add the folders Night_1, Night_2, ..., Night_n.

Weighted Batch Preprocessing

Note: It's important to separate keywords by an underscore _ and not by a simple dash - as otherwise the keyword will not be recognized by the script.

Run image integration for multi-session-images

The previous step creates one master light for each group (night_1, night_2, ...). If no groups were used the integration is now done but for multi-session images, we need to perform a separate image integration PROCESS > ImageIntegration > ImageIntegration.

Select Winzorized Sigma Clipping as Rejection algorithm under Pixel Rejection (1).

The input for this step is the registered frames that the WBPP script produced (located in the folder registered with the suffix _c_d_r.xisf). The stacked image will be displayed in PixInsight and should to be saved to have a starting point for future processings.

Prepare stretching

  1. Crop edges using PROCESS > Geometry > DynamicCrop.
  2. Remove gradients using PROCESS > BackgroundModelization > DynamicBackgroundExtraction.
  3. Color calibration using PROCESS > ColorCalibration > PhotometricColorCalibration.
  4. Noise reduction using PROCESS > MultiscaleProcessing > MultiscaleLinearTransform (optionally using a mask).


  1. Arcsinh stretch PROCESS > IntensityTransformations > ArcsinhStretch.
  2. Histogram transformation PROCESS > IntensityTransformations > HistogramTransformation.
  3. Remove green color cast PROCESS > NoiseReduction > SCNR.

Further processing

From here on the image processing differs from image to image so that there is no point in describing it. I will only point out some actions and processes and describe what they can be used for.

  • IMAGE > Color Spaces > Convert to Grayscale : Good starting point to create a mask of any sort.
  • PROCESS > Morphology > MorphologicalTransformation : Star reduction.
  • PROCESS > MaskGeneration > Starnet or PROCESS > <Etc> > StarNet2 : Separate stars from objects and create star masks.
  • PROCESS > Deconvolution > Deconvolution : Soften/Blur an image (useful for masks).
  • PROCESS > Convolution > Convolution : Sharpen an image.
  • PROCESS > IntensityTransformations > HistogramTransformation : Refining masks.
  • SCRIPT > Render > AnnotateImage : Create an annotated version of the image.

Full Screen Image