This page contains new experimental distributions of StarNet2 that haven’t been tested in full
CLI StarNet2 for MacOS with MPS support on M1 chips
This version of StarNet2 uses Torch and OpenCV instead of Tensorflow and LibTIFF. This required me to switch libraries around, but brought more possibilities. This version of the CLI tool has been fully reworked and supports many formats of image files as input and output, outputs mask, has an option for intermediate 2x upsampling. At this point the only difference between CLI and PI plugin is support for linear images (which depends on PI’s STF and can not be reproduced outside of PI at the moment).
This version contains signed binaries (so there should be no problems installing it) and should be blazingly fast on M1 CPUs!
Note: This is Command Line Interface (CLI) version, not PixInsight plugin!
Note 2: This version is currently compatible with SiriL v1.2 beta 2 and above!