Sep 24

My Processing Workflow

I was asked a few times to show how I process my images. So here is my (current) workflow which I use as basement. I created it over the last 2 years with many help from other people on AstroBin or tutorials in the internet.

At the bottom of this page, you’ll find my real check list, which I use for all my images – more or less, depending on the object (eg. Cygnus Loop needs star reduction twice or triple).

Note:

This is one way of many to process an image. Many processing steps can be done with other tools and maybe things can be done better than I’ll show here. But I don’t want spend too much time with processing. And I’m very happy with the results of this workflow every time 😉

Software used:
  • Pleiades Astrophoto PixInsight 1.8.5
  • Adobe Lightroom 6

 

Video Step-by-step

#1 Batch Preprocessing
#2 Image Crop
#3 Remove tint
#4 Background Extraction & Color Calibration
#5 Histogram
#6 Contrast
#7 Denoise
#8 Reducing Star Size
#9 Sharpness & Saturation
#10 Export
#11 Finish in Lightroom
Result
Check List

#1 Batch Preprocessing

I preprocess my pre-sorted raw images (bias, darks and lights) with the BatchPreprocessing script in PixInsight. Dependent on the object, I use 2x drizzle (especially for small objects).

In this example I don’t use drizzle, so I activate Image Integration in the Batch Preprocessing script.

Read more here.

On the right you see the result of the Batch Preprocessing. I already applied ScreenTransferFunction STF to stretch the image. Otherwise you see just a dark black image with some star points. Actually I don’t know why it has a very strong green tint…

Let’s start with the processing…


#2 Image Crop

First of all I crop the image. Because I want keep the image proportions, I use values at 4:3 ratio:

200px : 150px = 4 : 3

Note: values has to be negative!

If I want a square image or other formats, I use DynamicCrop.

 

Process – Geometry – Crop


#3 Remove tint

This part contains 4 steps:

ChannelExtraction

Channel Extraction splits the image into 3 channels. Red, Green and Blue.

 

Statistics

Statistics is used to find out which channel has the lowest median value, In this example it’s the red channel.

 

LinearFit

The channel with the lowest median value is used as reference image in LinearFit and applied to the other 2 channels (in my case G and B).

 

ChannelCombination

Final step of this part is ChannelCombination to combine R,G and B channel again to a RGB image.

 

Related video: Remove Tint In PixInsight

 

Process – ChannelManagement – ChannelExtraction
Process – ImageInspection – Statistics
Process – ColorCalibration – LinearFit
Process – ChannelManagement – ChannelCombination


#4 Background Extraction & Color Calibration

AutomaticBackgroundExtractor / DynamicBackgroundExtractor

To remove background gradients I use Automatic Background Extractor (ABE) or Dynamic Background Extractor (DBE), depending on the result with ABE.

ABE creates on some objects strange dark areas, well visible on M42 Orion Nebula:

AutomaticBackgroundExtractor
DynamicBackgroundExtractor

In that case DBE is much better to extract the background gradients. It is more time consuming but it is worth to spend the time. There are many good DBE tutorials on YouTube.

The next 3 steps are for calibrating the colors.

 

BackgroundNeutralization

As reference image I use a small star-less background area which is selected by using a preview mode (Image08_ABE->Preview01).

 

ColorCalibration

In ColorCalibration I use Preview01 as background reference and I create a new preview mode over the main object (Preview02) which I use as white reference. I didn’t adjust any other parameters here.

 

SCNR

Last but not least I remove slightly green noisy pixels (0.60-0.75) with a tool called SCNR (Subtractive Chromatic Noise Reduction). Read more about SCNR

 

Process – BackgroundModelization – AutomaticBackgroundExtractor
Process – ColorCalibration – BackgroundNeutralization
Process – ColorCalibration – ColorCalibration
Process – NoiseReduction – SCNR



#5 Histogram

All processes until now are applied to the unstretched raw image, temporary stretched with STF.

The unstretched histogram shows a very thin curve on the left side. We want wo stretch that thin curve and move it a little bit towards right.

 

Process – IntensityTransformations – HistogramTransformation

  1. F12 removes the ScreenTransferFunction STF
  2. Process – IntensiyTransformations – HistogramTransformation
  3. Adjust Shadows, but keep clipping value below 1000
  4. Adjust midtones to stretch the histogram curve

If needed, I adjust the midtones from single channels to reach best possible white balance.

 

Keep Shadows clipping value below 1000
Midtones adjustment to strecth the histogram
red channel adjustment



#6 Contrast

Curves Transformation

Here I use 2 points in RGB/K mode for slight contrast enhancement. One for the shadows in the left bottom corner (I drag the point a little bit downwards) and a second point in the center of the curve (I drag the point a bit towards left).

Not too much, final contrast can be adjusted later in Lightroom or of course also in PixInsight.

 

Process – IntensityTransformations – Curves



#7 Denoise

As denoise tool I use TGV Denoise. In the past I used ACDNR. Until I found a comparison of different denoise algorithms. And TGVDenoise performed best.

Because I want only denoise the dark regions I use a mask.

 

4 steps to denoise the image:
  1. First I create a CIE L component mask
  2. Next the CIE L component mask is beeing inverted
  3. Then the mask is applied
  4. Finally I use TGVDenoise with strength greater than 8.0 and smoothness greater than 5.0

 

Extract CIE L component
Invert extracted CIE L component
Add mask
Process – NoiseReduction – TGVDenoise



#8 Reducing Star Size

This is a important step to give the main object more presence.

  1. With StarMask I create a mask, usually with the default parameters.
    Only if I process a drizzled image I use Scale: 6
  2. MorphologicalTransformation reduces the star size. Here I play with the amount value.
  3. If needed, I repeat this 2 steps to reduce the stars more and more…

 

Related video: Reduce Star Size In PixInsight

 

Process – MaskGeneration – StarMask
Process – Morphology – MorphologicalTransformation



#9 Sharpness & Saturation

Final steps in PixInsight…

Again I use a mask: RangeSelection mask.

Alternative: If the object contains only one color, ColorMask is a very good alternative (Script – Utilities – ColorMask).

 

Steps
  1. Create RangeSelection mask
  2. Adjust sharpness with UnsharpMask
  3. Saturation anhancement with CurvesTransformation

Sometimes I invert the RangeSelection mask at the end, to de-saturate only the background (again with CurvesTransformations).

 

Process – MaskGeneration – RangeMask
Process – Convolution – UnsharpMask
Process – IntensityTransformations – CuvesTransformations



#10 Export

Last step in PixInsight is export the final image as 16 bit TIFF, without associated alpha channel.

If I wouldn’t use Lightroom for the finish, I would now enhance the contrast in PixInsight with CurvesTransformations (2 points for contrast).

 




#11 Finish in Lightroom

As last step I import the 16 bit TIFF in Lightroom. If needed I crop the image again. Then I start to Develop the image in the following order:

Section: Lens Correction / Detail

Remove Chromatic Aberration, Color Noise Reduction and Luminance Noise Reduction

Section: Basic

Exposure, Highlights, Shadows, Blacks, Clarity, Vibrance and Saturation

Section: Tone Curve

Darks and Shadows

Attention: clipping!

Take care to the histogram. If the little triangles on the top left and right corner in the Histogram changes the color from grey to any other color, it’s clipping.

 




Done 🙂

The result of this quick processing. On my original NGC7000 image I used additionally HSL-Hue in Lightroom to adjust red/yellow colors.

So, I hope this was helpful… definitely it was a lot of work to create this tutorial 😉


Check list

  • 1. Crop
    2. ChannelExtraction – LinearFit – ChannelCombination (Global Apply)
    3. AutomaticBackgroundExtractor (2x ?) / (Dynamic Background Extractor)
    4. (BackgroundNeutralization) – ColorCalibration
    5. SCNR to green 0.60-0.75
    6. Histogram
    7. If needed: Curves with 2 points! (enhance contrast)
    8. Extract CIE L component mask (inverted) – TGVDenoise
    9. StarMask (Drizzle: scale 6 for larger stars) – MorphologicalTransformation (2x ?)
    10. MaskGeneration (Range) – UnsharpMask – (LocalHistogramEqualization) – CurvesTransformation
    11. if needed: without any mask – CuvesTransformation – Saturation
    12. if needed: again StarMask – MorphologicalTransformation
    13. if needed: ColorMask – CurvesTransformation (R/G/B, RGB, Ciel, Sat.)

 

8 comments

  1. Rainer Späni

    Wow, very helpful. Thanks a lot for your effort.
    I have to a lot to learn…

  2. Jan Schubert

    Großartig und vorallem verständlich! Danke für die Mühe, Karol!
    Gruß Jan

  3. Matthias

    Great job. Well done!

  4. Corey

    Thank you so much for making this turtorial! Your work is so inspiring, and you sharing your process is greatly appreciated! Thank you thank you thank you!

Leave a Reply

*

%d bloggers like this: