© astropix.nl
But the (now well-rested) amateur is still not satisfied, and turns his attention to the final frontier: deconvolution. An imaging instrument like a telescope does not image a point source (like a star) as a point, but as a sizeable blob with a diameter proportional to the brightness of the source. The size of the blob is increased by atmospheric turbulence. Obviously, the image quality will be improved if these blobs can be removed somehow, i.e. deconvolved.
The PixInsight image processing software https://pixinsight.com/ does this by sampling the blobs of unsaturated stars in the image, and creating a blob model from that information. This model is then removed from the image at the position of the various blobs. This is a laborious task, but absolutely worth the time. During deconvolution, the sky background is protected by a mask and after the process has finished one is rewarded with a better looking image (*).
The above image of a part of the Hercules galaxy cluster was deconvolved, at left the original, at right the final version. The improvements are most noticeable in the extended objects. The full size deconvolved image can be viewed here:
https://www.astrobin.com/full/329643/0/?nc=user
(*) The difference between the two images is fairly small, because the Point Spread Function (blob shape) is small in the case of the fully sampled aperture plane of a typical optical telescope. The effects of (de)convolution are much more dramatic in radio interferometry (ASTRON's core business), since the PSF of a radio aperture synthesis telescope is much more extended due to sparse sampling of the aperture plane.