So due to a good breakthrough with fluid.jit.plotter (via some new info/knowledge from the Max beta forum being passed to @balintlaczko) I’ve been resisting 2d plotting of datasets for corpus scrubbing ala CataRT et al.
At the moment I’m doing something like this (ala my old experiments with fluid.datasetplot~):
This is really cool/handy for using a mouse, but I’d like to be able to navigate this with a gamepad or other grid-y surfaces.
Sadly fluid.grid~ absolutely chokes on a dataset of any significant size (as it turns out!). The dataset on the right is ~30k points and that pinwheeled Max for like 5min before I decided to just force quit it.
I imagine it’s doing loads of stuff trying to optimize the relationships between all the points, but for larger data, it seems like that’s not really viable. And there’s not really a lot of parameters to tweak with the algorithm to try and make it faster/dumber, so I’m wondering if there’s a cheap/fast way to make something grid-y out of larger datasets like this. Like some kind of brutal ordering/quantization or something like that? And/or is there any settings or pre-processing that can make fluid.grid~ happy with 30k data points.
