My PixInsight Processing Workflow for Nebulae
Raw astrophotography data looks nothing like the final image. Here is the step-by-step PixInsight workflow I use to bring nebulae to life.
The most common question I get is: "How do you process your images?" It is a fair question. Raw astrophotography data looks nothing like the finished picture. A freshly stacked frame is dim, gray, and flat. The nebula is barely visible. The transformation from raw stack to finished image is where the real craft lives.
I process everything in PixInsight. It has a steep learning curve, but once you understand the underlying logic it is the most powerful tool available for this work. Here is my current workflow for narrowband nebula images.
Step 1: Calibration
Before stacking, every light frame needs to be calibrated to remove sensor artifacts. I use three types of calibration frames:
- Bias frames — zero-second exposures that capture the sensor's read noise baseline
- Dark frames — same temperature and duration as lights, capturing thermal noise
- Flat frames — evenly illuminated frames that map the optical train's vignetting and dust
I use the ImageCalibration process in PixInsight, applying master bias, dark, and flat frames created from 30+ individual calibration frames each. Good calibration is the foundation of everything that follows.
Step 2: Stacking
With calibrated frames in hand, I use StarAlignment to register all frames to a reference, then ImageIntegration to stack them. My typical settings:
- Combination: Average
- Rejection: Winsorized Sigma Clipping (excellent for moderate frame counts)
- Weights: Noise evaluation (lets PixInsight weight better frames more heavily)
For narrowband work I stack each channel (Ha, OIII, SII) separately, producing three master stacks.
Step 3: Background Extraction
Fresh stacks almost always have gradient — uneven background illumination from light pollution, moonlight, or optical vignetting that survived the flat calibration. I use DynamicBackgroundExtraction (DBE) to model and subtract this gradient.
The key is placing sample points only on true background sky — not on nebulosity, not on bright stars. I take my time here. A poorly placed DBE can destroy faint outer structure that took hours to collect.
Step 4: Star Removal
This is a relatively recent addition to my workflow, but it has transformed my results. I use StarNet2 to generate a starless version of each channel. Processing the nebula without stars means I can apply aggressive noise reduction and stretching without worrying about bloating or halos around bright stars.
I keep both the starless and star-only versions and recombine them at the end.
Step 5: Noise Reduction (Linear)
Before stretching, I apply NoiseXTerminator to the linear (unstretched) starless frames. Noise reduction in the linear state is more effective than in the stretched state because the signal-to-noise ratio is more uniform across the frame.
I use conservative settings here — just enough to smooth the background without softening real nebula structure.
Step 6: Stretching
This is the most visually dramatic step. The linear stack is stretched from a nearly black frame to a full-range image using HistogramTransformation or the Generalized Hyperbolic Stretch (GHS) script.
I prefer GHS for nebulae because it gives fine control over how the stretch affects different parts of the tonal range. I can protect the faint outer wisps while preventing the bright core from clipping — something that is difficult to achieve with a simple histogram stretch.
Step 7: Color Combination
For HOO images I use ChannelCombination to assign Ha to Red and OIII to both Green and Blue. For SHO (Hubble Palette) I assign SII→R, Ha→G, OIII→B.
The initial color combination almost always looks garish — oversaturated greens and magentas. This is normal. The next step fixes it.
Step 8: Color Calibration and Adjustment
I use PhotometricColorCalibration (PCC) to set a proper white balance, then CurvesTransformation to adjust individual color channels. For HOO images I typically:
- Boost the red channel slightly to warm the Ha emission
- Shift the green-blue balance to give OIII regions a cooler, more distinct teal tone
- Reduce overall saturation slightly for a more natural look
This is the most subjective part of the workflow. I am making artistic decisions about how I want the final image to feel.
Step 9: Final Sharpening and Detail Enhancement
With the color image looking good, I apply UnsharpMask or LocalHistogramEqualization selectively to the brightest nebula structures. The goal is to enhance fine detail — the pillars, the filaments, the turbulent edges — without introducing artifacts.
Step 10: Star Reintegration
Finally, I add the stars back using PixelMath: max(starless_image, star_layer). I often reduce the star layer's brightness slightly before combining so the stars do not overwhelm the nebula. The result is a natural-looking image where the nebula is the clear subject.
The Time Investment
A typical narrowband image takes me 4–8 hours of processing time spread over several sessions. I rarely process an image in one sitting — I find it helps to step away and come back with fresh eyes, especially for the color and curve work.
The processing is as much a part of the craft as the imaging. Every decision shapes the final result. If you are just getting started with PixInsight, be patient with yourself — the learning curve is real, but the results are worth it.
Clear skies.
Explore Topics
Written by
Joseph S. Volpe
Content creator and writer sharing insights and stories.