Bi-colour narrowband imaging using H-alpha and OIII filters is one of the most accessible ways to produce striking deep-sky images — and it works well even from light-polluted locations. You only need two filters, and the processing workflow is straightforward once you understand the steps.

This guide walks through the complete process from calibrated frames to a finished colour image. It assumes you have already captured your Ha and OIII sub-exposures and have matching calibration frames. If you are unfamiliar with narrowband filters, read the Narrowband Filters Explained guide first.

What you need before starting

  • Ha sub-exposures: 20+ frames at 3–10 minutes each (more is better)
  • OIII sub-exposures: 20+ frames at 5–15 minutes each
  • Calibration frames: darks, flats, and (optionally) bias frames for each filter. See the Calibration Frames guide for details.
  • Stacking software: Siril, DeepSkyStacker, PixInsight, or equivalent
  • Post-processing software: Siril, PixInsight, Photoshop, GIMP, or equivalent

Step 1: Calibrate and stack each channel separately

Process Ha and OIII as two completely independent image sets.

For each channel:

  1. Apply calibration: Subtract master dark, divide by master flat, subtract bias (if using)
  2. Register (align): Use star-based registration to align all sub-exposures
  3. Stack: Use sigma-clipping or winsorised sigma-clipping rejection to combine frames and reject outliers (satellites, cosmic rays, etc.)
  4. Save the result as a 32-bit or 16-bit linear FITS or TIFF

You should now have two stacked master images:

  • master_Ha.fit (or .tif)
  • master_OIII.fit (or .tif)

Sanity check: Open each master and verify the nebula is visible, stars are round, and there are no stacking artefacts.

Step 2: Align the two masters to each other

Your Ha and OIII stacks may not be perfectly aligned (especially if you used different imaging sessions or moved the camera between filter changes).

  1. Use star-based registration to align master_OIII to master_Ha (or vice versa)
  2. Crop both images to the common overlap area

Step 3: Create the HOO colour composite

The HOO (Ha-OIII-OIII) mapping assigns:

Colour channel Data source
Red Ha
Green OIII
Blue OIII

In Siril:

  1. Open the Ha and OIII masters
  2. Use Image → Colour Compose (or the rgbcomp command)
  3. Assign: R = Ha, G = OIII, B = OIII
  4. Save the resulting colour image

In PixInsight:

  1. Use ChannelCombination with R = Ha, G = OIII, B = OIII
  2. Apply to create a new colour image

In Photoshop/GIMP:

  1. Open both masters as greyscale images
  2. Create a new RGB document
  3. Paste Ha into the Red channel
  4. Paste OIII into both the Green and Blue channels
  5. Flatten

Step 4: Stretch the image (linear to non-linear)

Your composite is still in a linear state — most of the data is very dark. You need to stretch it to reveal detail.

Histogram stretch approach:

  1. Apply a gentle stretch using histogram transformation (Siril: Asinh Stretch or Histogram Transform; PixInsight: ScreenTransferFunction + HistogramTransformation)
  2. Bring up the midtones gradually — do not clip the highlights or crush the blacks
  3. Aim for a result where nebula detail is clearly visible but not over-stretched

Tips:

  • Stretch each channel equally for a balanced colour result
  • If the image looks too green (OIII dominates), you can reduce the green contribution slightly
  • If the image looks too red (Ha dominates), reduce the red contribution slightly

Step 5: Colour balance and saturation

After stretching, the colours often need refinement.

Common adjustments:

  • Colour balance: Shift the background toward neutral grey/black. A faint green or magenta cast is common and should be removed.
  • Saturation: Increase saturation moderately to bring out the colour contrast between Ha (warm red/orange) and OIII (cool teal/blue) regions.
  • Star colour: Stars in narrowband images are often oddly coloured (purple, green, or white). You can:
    • Desaturate stars using a star mask
    • Replace star colour with broadband RGB star data if you have it
    • Accept the false-colour star appearance

Step 6: Detail enhancement

With the colour composite looking good, enhance fine detail:

  1. Sharpening: Apply a gentle unsharp mask or deconvolution to bring out filamentary structure. Be conservative — noise amplification is the enemy.
  2. Noise reduction: Apply luminance noise reduction to smooth the background without smearing nebula detail. Masks help protect bright regions.
  3. Local contrast: A mild curves adjustment targeting the midtones can add depth to the nebula structure.

Step 7: Final adjustments and export

  1. Crop to remove any edge artefacts from registration
  2. Set the background to a neutral dark grey (not pure black — keep some natural background)
  3. Adjust overall brightness and contrast to taste
  4. Export as JPEG (for web/sharing) and TIFF (for archival)

Expected results

A well-processed HOO image will show:

  • Warm red/orange regions where hydrogen emission dominates
  • Cool blue/teal regions where oxygen emission is strong
  • Transition zones with blended colour
  • Fine filamentary and structural detail in the nebula

The result is visually striking and scientifically meaningful — each colour represents a different physical process in the nebula.

Common mistakes

Mistake Fix
Channels not aligned Register masters to each other before combining
Over-stretched background Use a gentle, iterative stretch
Green stars Apply star desaturation or use a star mask
Noisy OIII channel Capture more OIII sub-exposures (OIII is often fainter than Ha)
Unbalanced colour Adjust channel weights before or after stretching

FP Softlab context

The FP Softlab deep-sky gallery features narrowband nebula imagery. The Cygnus NGC 7000 mosaic is a hydrogen-alpha target that demonstrates the kind of structure Hα imaging reveals.

Further reading