Overview

The Stacker module handles the alignment and stacking of calibrated subs.

Configuration

Control

The Stack module is launched in the background at ALS startup

Input

Behavior

flowchart LR
Start([START])
FirstSub{{First sub of the session?}}
SetAlignReference[Set sub as alignment reference]
CheckShape{{Sub same dimensions as previous result?}}
CheckAlign{{Alignment ON?}}
AlignImage[Align sub]
StackImage[Add sub to Stack]
ComputeStacking[Compute stacking]
PublishReference[Return alignment reference]
PublishResult[Return generated image]
End([END])

Start --> FirstSub
FirstSub -- YES --> SetAlignReference
SetAlignReference --> PublishReference
FirstSub -- NO --> CheckShape
CheckShape -- YES --> CheckAlign
CheckAlign -- YES --> AlignImage
AlignImage --> StackImage
CheckAlign -- NO --> StackImage
StackImage --> ComputeStacking
ComputeStacking --> PublishResult
CheckShape -- NO --> End
PublishReference --> End
PublishResult --> End

classDef bounds fill: #333, stroke: #666, stroke-width: 2px, color: #BBB, font-family: 'Poppins', sans-serif
classDef step fill: #444, stroke: #622, stroke-width:2px, color: #c6c6c6, font-family: 'Poppins',sans-serif
classDef test fill: #444, stroke: #226, stroke-width: 2px, color: #c6c6c6, font-family: 'Poppins', sans-serif

class Start,End bounds
class SetAlignReference,AlignImage,StackImage,PublishResult,ComputeStacking,PublishReference step
class CheckShape,CheckAlign,FirstSub test

Alignment

If alignment is ON

1. Search for similarities between the calibrated sub and the session alignment reference.

   If the calibrated sub has similarities count below the configured detection threshold, it is discarded and the Stack module resumes listening to its queue.

2. Compute required transformations for the calibrated sub to align with the reference:

   • translations
   • rotation
   • resizing

3. Apply the transformations to the calibrated sub.

Stacking

1. Add the aligned (if requested) sub to the stack.
2. Generate a new image containing the stacking result according to the configured mode.

When working in mean mode and the current profile is Astrophoto, ALS automatically removes transient bright artefacts such as satellite trails by detecting and clipping per-pixel outliers.

The stacker keeps a Welford-based running mean and variance for every pixel, and once at least 3 frames have been accumulated it clips any new pixel sample that lies above the previous mean plus 4σ.

Clipped values are replaced by the previous mean, providing single-pass, per-frame rejection without extra iterations.

Output

The generated image is broadcast