A few frontiers I find genuinely exciting:
The time dimension. At the fast end: photon-counting detectors (SPADs), smart-telescope networks and robotic rigs are opening true time-domain work to amateurs. Occultations, transits, variable stars, the optical counterpart of a transient alert. At the slow end: multi-year and multi-decade animations. The classic example is the Crab Nebula, whose expansion is now measurable from amateur data taken years apart. (Given our name, you'll forgive me a particular soft spot for the whole time-domain idea.)
LEO. It sounds exotic, but rideshare launch costs and miniaturised optics are making it less fanciful every year. The first genuinely amateur-grade orbital imager feels like a "when," not an "if", and it would reset what "good detail" even means.
A wider gamut. We're still mostly living inside the LRGB and SHO color spaces, and fairly narrow gamut JPGs by any measure. Finer filters decomposition would let us map the sky into a far larger, more truthful color space. There's a lot of real color information out there we're simply not recording (or displaying) yet.
High dynamic range. Not bracket-stitching, but true HDR captured natively. Think the Trapezium core and the faintest outer shells of M42 in a single self-consistent dataset. Photon-counting and high-well/log sensors, as well as higher dynamic range monitors point in this direction.
Teams. Remote scopes make 100+ hour integrations routine, but pooled, collaborative datasets has proven to take us to depths no single imager reaches : ultra-faint IFN, tidal streams, planetary-nebula halos.
Smarter processing. Sub-frame deconvolution, PSF modelling and ML can pull out real detail. The genuine challenge that comes with them is provenance : the line between enhancement and fabrication will define good practice for the next decade.
In short, I don't think we're done. The polished, repeatable image is becoming a baseline. It frees our attention for the harder, stranger, more original work.
Curious to hear where others think the real edges are.