It would be great to have more Julia packages for downsampling huge datasets for visualization, since there are a variety of strategies for different kinds of data.
For inspiration, see e.g. the Python datashader package for visualizing point clouds, this thesis on downsampling huge timeseries, this paper on downsampling volumetric data giving a “new low-cost downsampling filter” (and also nicely reviewing a lot of related algorithms), and the Julia GigaSOM.jl package for visualizing huge cytometry datasets. There is also this discussion thread on software for visualizing huge medical images (hundreds of GiB).