Description
Stacking misalignment (failed registration) occurs when the stacking software fails to correctly align frames before combining them.
The typical symptom: doubled, tripled, or trailed stars, sometimes localized to one region of the field. Instead of a single point, each star appears multiple times, offset from itself.
Causes vary: insufficient star detection (sparse field, noisy frames), field rotation between frames, mixing of flipped frames or frames from different focal lengths, or an inappropriate alignment model.
This is a processing defect, correctable by re-stacking. It is related to field rotation and tracking drift when the root cause is in acquisition rather than the software.
Visual signature
Stars appear doubled or multiplied: two, three, or more slightly offset points where there should be only one.
The offset may be uniform (global misalignment) or increasing toward the edges and arc-shaped (uncompensated field rotation at stacking).
In partial cases, only a subset of frames is misaligned, producing stars with a weaker ghost replica beside the primary point.
Unlike tracking defects (present on each individual frame), the individual subs are sharp here: it is the combination step that introduces the doubling.
Differential diagnosis
Distinguish from field rotation: field rotation already deforms each individual frame into arc-shaped stars, whereas stacking misalignment can start from sharp subs and only appear after stacking. Inspect an individual sub to decide.
Separate from tracking drift and periodic error: those defects elongate stars on every individual frame, while misalignment doubles them (a clean, offset replica) rather than producing a simple trail.
Do not confuse with an internal reflection: a ghost is a replica of a bright source, not a doubling of every star in the field.
Decisive test: if the individual subs are sharp but the master shows multiple stars, the problem lies in the stacking step.
Probable causes
- Insufficient star detection (sparse field, noisy frames)
- Field rotation between frames not compensated at stacking
- Mixing of flipped frames (meridian flip) without re-alignment
- Inappropriate alignment model (translation only instead of including rotation)
- Mixing frames from different focal lengths, sensors, or image scales
- A few heavily misregistered subs corrupting the global registration
Course of action
- Check an individual sub first to confirm the problem is in the software
- Improve star detection by lowering the detection threshold on a sparse field
- Choose an alignment model that handles rotation and distortion
- Properly re-align frames from both sides of a meridian flip
- Exclude the few subs that are breaking the registration
- Stack with rotation cropping when genuine field rotation is present
- Do not mix frames from different image scales or sensors without resampling
The Doc's advice
Doubling after stacking is almost always fixable without collecting a single new frame: the problem is in the stacking step, not in your subs. First, check an individual sub -- if it is sharp, the registration is to blame. On the software side, make sure star detection finds enough reference points (lower the threshold on a sparse field), do not stack frames from before and after a meridian flip without re-alignment, and choose an alignment model that handles rotation if your frames drift rotationally. If a small number of subs is breaking the registration, exclude them rather than degrading the entire master.
Think you can see this defect in your image?
Run a diagnosisFrequently asked questions
Why are my stars doubled after stacking?
Because registration (the alignment of frames) has failed. Before combining images, the software must precisely overlay the stars in each frame; if it cannot do that, the frames are offset from one another and each star appears doubled or tripled. Proof that the problem is in stacking and not in acquisition: your individual subs are sharp. The usual causes are insufficient star detection, rotation between frames, or mixing of flipped frames.
How do I fix a stacking misalignment?
By re-stacking with better settings. First check that an individual sub is sharp (if not, the problem is in acquisition). Improve star detection by lowering the threshold on a sparse field, choose an alignment model that can handle rotation and distortion (not only translation), and make sure frames from both sides of a meridian flip are properly re-aligned. Finally, exclude any subs that are corrupting the registration rather than degrading the entire master.
Does the doubling come from acquisition or from processing?
The test is simple and decisive: open an individual sub. If it shows a single, sharp star, the doubling appeared during stacking (failed registration) -- a processing problem that can be fixed without retaking any data. If the stars are already doubled, elongated, or arc-shaped on the individual sub, the problem is in acquisition (tracking drift, field rotation, vibration). This check immediately points you toward the right family of solutions.
Can frames taken on both sides of the meridian be stacked together?
Yes, provided they are properly re-aligned. During a meridian flip, the sensor ends up rotated 180 degrees relative to the sky: frames taken before and after the flip are therefore inverted relative to each other. Most stacking applications handle this automatically if the alignment model allows rotation, but you need to confirm that is actually the case. Otherwise, a characteristic doubling results. If in doubt, process the two groups separately and then combine the two re-aligned masters.