Cookie consent

AstroBin saves small pieces of text information (cookies) on your device in order to deliver better content and for statistical purposes. You can disable the usage of cookies by changing the settings of your browser. By browsing AstroBin without changing the browser settings, you grant us permission to store that information on your device.

I agree
Contains:  Other
StarNet: an update, 


            Nikita Misiura


This is an update for my StarNet project: a neural network to remove stars from images. Previous releases can be found here, here and here.

This release includes two major changes:

1. There are now two executables: for Grayscale images and for RGB images. Both still work with 16bit per channel TIF images. Previous release was created with only color images in mind and performance on Grayscale images was often rather poor. This should not be the case anymore.

2. Both Windows and MacOS are now supported (make sure you download proper version).

UPDATE: Some people report troubles running the software on Mac. If you do, try to folllow instructions here, or let me know.

The only thing I ask is that if you use the program, then add it into the list of software you used. Thank you.

For some technical notes please read description here.

Important usage notes:

1. It is usually better to use it on images with original resolution. On heavily downsampled images it may leave the smallest stars untouched (like some sort of a "dust").

2. Very dark regions of images may be problematic and leave more notable artefacts. It might be better to push up dark regions before feeding it to the program.

3. The program is designed to work on minimally processed images with undistorted star profiles. Ideally you should just stretch your data, use starnet to remove stars, and then proceed with further processing. Remember that it might not work very well on that random image you just found on the internet, good old "garbage in -> garbage out" principle still applies. Star profiles should not be distorted too much (due to optical imperfections or heavy processing) for it to work well, otherwise it might miss some stars or leave artifacts. These are not hard rules, however, so you can try to use it on any images you want, just please balance your expectations and be aware of limitations.

Hope it works for you, if not, please let me know! All feedback is welcome!



PS: Git repo will be updated a bit later! I am not sure how to handle two versions (RGB and Grayscale).



Nikita Misiura
License: Attribution-NonCommercial Creative Commons


StarNet: an update, 


            Nikita Misiura