v0.1.1

First packaged release that actually runs end-to-end. Onboarding goes from a 2-hour native install to a single docker run.

Quick start

# CPU
docker run -d -p 8000:8000 \
  -v ~/Pictures/Siril:/captures:ro -v astrolab:/data \
  ghcr.io/bscholer/astrolab:latest

# NVIDIA GPU
docker run -d --gpus all -p 8000:8000 \
  -v ~/Pictures/Siril:/captures:ro -v astrolab:/data \
  -e ASTROLAB_ENABLE_STARNET=1 \
  ghcr.io/bscholer/astrolab:cuda

Open http://localhost:8000, go to Settings, set the captures path to /captures, hit Refresh, and your sessions appear.

What's in the image

• Siril 1.4.2 (GPLv3) extracted at /opt/siril
• GraXpert 3.0.2 (GPLv3) at /opt/graxpert
• StarNet++ v2 (proprietary freeware) fetched on first run when ASTROLAB_ENABLE_STARNET=1
• cuDNN 8.9 layered on the CUDA image for GraXpert's onnxruntime
• The astrolab API and SvelteKit UI served on a single port (:8000)

Sizes: CPU 2.82 GB, CUDA 8.26 GB.

Tags

Tag	Means
:latest	most recent CPU build from main
:cpu	same as :latest
:cuda	most recent CUDA build from main
:v0.1.1	this release, CPU
:v0.1.1-cpu	explicit alias
:v0.1.1-cuda	this release, CUDA

Credit

astrolab is an orchestration layer. The real image processing is done by Siril, GraXpert, and StarNet++ v2. The OSC stacking pipeline follows Naztronomy's smart-telescope script.

Heads up

v0.1.0 shipped with a broken calibrate_register_stack template (KeyError: rgb_equal at job build). Pull v0.1.1 instead, or just :latest / :cpu / :cuda. v0.1.0 is left in place for posterity but should not be used.

Status

This is a 0.x release. Schemas, template versions, and cache keys can still change between commits; a project from today is not guaranteed to open cleanly in v0.2. The V1 milestone tracks the punch list.