Actually, when AI Denoising existing existing images, SeeStar reads the JPG, not the FIT file! This makes absolutely no sense, why denoise a heavily compressed and much noisier JPG image when you could start with the much cleaner and higher dynamic range FIT file? And not only is this JPG only, but the images are watermarked, which means you lose the bottom 10% of the image.
How difficult can it be to replace JPG read and write image function calls with FIT ones? This is such a low hanging fruit it's basically sitting on the ground.
