Align Channel Masters with StarAlignment

Channel combination and image blending expect corresponding pixels to represent the same sky position. Masters integrated separately across filters or sessions can differ in framing, rotation, or image scale. StarAlignment uses stars to register targets to a reference image’s grid.

If WBPP already registered all channels to the same reference and the masters line up, verify that result and skip another resampling. This tutorial handles integrated light masters that need alignment; it does not replace registration of individual exposures before stacking.

Work after integration and before a common crop, gradient correction, and channel combination. Keep the pixels linear and stars present. Registration changes geometry, not color balance, background gradients, or relative intensity matching.

Save the original integrated light masters from WBPP and work on copies. Use actual filter names such as Target_R, Target_G, Target_B or Target_Ha and Target_OIII. Do not register darks, flats, or bias masters to the sky reference.

Check that every input shows the same field with substantial shared coverage and recognizable stars. Use STF for inspection without permanently stretching. A raw color-camera mosaic should already have been calibrated, debayered, and integrated for this master-level exercise.

Different dimensions or orientation can be legitimate reasons to register. Missing field coverage, severe blur, and very different star profiles are limitations to inspect, not problems solved merely by matching image dimensions.

Choose a clean master with reliable stars, useful field coverage, and an appropriate image scale for the intended output. A good luminance or broadband master can be convenient; a well-exposed narrowband master can also work. Do not choose solely by which STF looks brightest.

Keep this reference unchanged. Align every other target directly to it rather than chaining one transformed channel to the next. Keep a copy of the reference alongside the registered outputs so the final set has one clear geometric baseline.

For different cameras or telescopes, consider the final field and sampling deliberately. Registering coarse data to a fine grid does not create captured detail, and a large resolution mismatch makes this a more demanding exercise. Begin with compatible same-field data before tackling those cases.

Find StarAlignment through All Processes, usually under ImageRegistration. Reset a reused instance and consult the installed help for current models and defaults.

Use Blink or matched views at 1:1. Compare star centers near the middle and all corners. Filters naturally show different star brightness and object structure; look for position changes rather than demanding identical appearance.

Check output dimensions against the reference and inspect borders for uncovered areas. Matching dimensions alone does not establish alignment. Conversely, black output borders can be expected where a target lacks coverage of the reference field.

Look for corner offsets even when the center looks good. Those can indicate distortion or a registration-model mismatch. Severe star-shape differences can also produce color rims later even when centroids align; inspect focus and optical quality separately.

Do not use a common crop to conceal registration errors inside the useful field. Accept the geometry first, then remove invalid coverage consistently in the next lesson.

Use the same unchanged reference and validated configuration for the remaining targets. Preserve per-filter identities and check that every intended target has a successful output. Do not accidentally include an already registered result in another pass.

The reference itself can remain as its original saved copy; it does not need another registration merely to obtain the output postfix. Keep a checklist identifying that reference and each registered companion.

For RGB, the set is the reference plus the aligned R/G/B companions as appropriate. Include L if needed for LRGB, or an RGB integration if adding its stars to a narrowband image. For HOO, align the acquired Hα and O III masters; reuse O III later rather than generating separate green and blue registrations.

First check the reference and target paths, field overlap, and whether recognizable unsaturated stars exist in both. Very weak narrowband data, hot pixels, severe gradients, or poor focus can make star detection unreliable.

Use a detected-stars diagnostic mode where available to inspect what is being identified. Return to the ordinary registration mode afterward. Change detection settings one at a time and test a single target; do not relax every control until an arbitrary match succeeds.

If the center aligns but edges drift, consult the official distortion tutorial and current process help before changing registration models or enabling distortion correction. A more flexible fit needs validation at many star positions; it is not automatically better.

A failed target may need improved input data or a more specialized registration approach. This beginner workflow is not a mosaic-merging recipe. Do not force images with little overlap into a plausible-looking color combination.

Save the validated aligned masters and identify the usable intersection across the entire set. Follow the DynamicCrop lesson to apply one common crop, then proceed to gradient correction and optional LinearFit before combination.

Recheck astrometric metadata after registration and cropping. A geometric match does not itself establish a valid sky solution for MGC or SPCC. Establish the required solution and fresh flux-calibration metadata on the actual current working image when those processes require them.

Rebuild or verify masks and saved sample coordinates tied to earlier geometry. Record the reference, input paths, output paths, registration settings, and field checks so future channels can be brought onto the same intended grid.

This is an original practice sequence using established registration workflows. Linked examples include older interfaces; installed StarAlignment documentation governs current controls. Keep all unregistered originals for future experiments without repeated interpolation.