Russell Croman, the developer of the XTerminator line of software, has released version 2.0 of the popular and amazing StarXTerminator along with three new AI models for various applications.

The new version is a significant improvement over the previous version; there are fewer artifacts, even in JWST images that have large diffraction spikes on many of the stars.

AI version 11 is an all-new neural network that’s been retrained from scratch. The neural network comes in three types – use the one that best suits your needs – here’s an overview of what each type of neural net does:

  • The full version produces the best results but uses the most compute power; this version will run the longest and may not even run on older hardware.
  • The ‘Lite’ version uses less memory and compute power and might be suitable for older systems.
  • The ‘Lite-no-noise’ version requires the least compute power and memory usage but does away with the noise matching process (if you plan to use NoiseXTerminator on your image, you can use this version since noise matching would be unnecessary because you’re removing the noise later anyway).

The following image is NGC 3324 as imaged by the JWST and processed by myself:

There are stars of various types and sizes, including stars with large diffraction spikes on them. This is the image after running StarXTerminator version 2.0 with AI 11 (full version):

The image is very smooth and even the stars with the diffraction spikes have been removed. StarXTermaintor correctly identified the galaxies in the image and did not modify them.

If you’d like to see a video where Croman is interviewed and goes into detail about the new version and new AI model, check this video: https://www.youtube.com/watch?v=ZuNQG1WFeGs.

If you haven’t tried StarXTerminator, Croman makes a free 30-day trial available, and I highly recommend that you add it to your workflow.