An extremely casual 10mins on the Horsehead two nights ago. I set up in nice seeing, started an image on an LMC object juuust as the clouds rolled in... so I slewed away from the clouds and got this instead, before the clouds got everything. 5x2min hydrogen alpha, C14 at f/6, Atik 460EX, good seeing, but no calibration frames as I wasn't expecting anything worth processing! Processed with PixInsight, including RCAstro's BXT, NXT & SXT. Stars were decently small and round even before the rather impressive magic of BXT.
As an aside, I have recently tried and been very impressed by these RCAstro machine-learning tools - as a planetary imager I familiar with what good deconvolution can do to decent or even ordinary data, it can be extraordinary. Also just how much more potential resolution there perhaps is in deep-sky data (overlay a hi-res planet on your favourite nebula at the same scale and you'll see what I mean, even a fraction of that is a massive improvement!). The tools work well and make for a pleasing result - and as I hadn't fully got my head around the inner workings of deep-sky decon methods this has been a joy to play with as it is a lot easier to run. I think my slightly oversampled C14 data works rather well with the ML tools here. I have seen a couple of instances where the tools produce errors, such as a faint galaxy being turned into two stars (maybe showing for sure it's not "cheating"), but mostly the results are faithful to my data. I was initially a bit suspicious, but am now a convert to these tools, if carefully used.
Higher res link