wire mother @echo.northsky.social · 7h"iny glass house throwing stones* GO FUCK YOURSELF GEORGE KELLING 000
wire mother @echo.northsky.social · 8hit's so fucking annoying that any time you try look for papers or whatever about calculus/algebra it's all llm stuff. like, no im not actually trying to learn about anything above 5 or 6 dimensions thanks. auto social engineering malware sucks 110
wire mother @echo.northsky.social · 8hcorrection: linear calculus is boring. i could fw complex calculus. i WILL learn your ways mr clifford 000
wire mother @echo.northsky.social · 8hshould have paid more attention in maths class. abstract algebra is rad, but calculus is kinda boring 100
wire mother @echo.northsky.social · 8hone day I'll figure gf2 coz i just know there's gotta be a non-cryptography use for it like smth with raster bitmaps. maybe some kind of binary NURBS or smth 100
wire mother @echo.northsky.social · 8hmy downfall will probably be my aversion to using any code I don't understand. i love a succinct algorithm, but will spend lots of time figuring out exactly how it works to the point i can reproduce it from first principles. unfortunately i love theory much more than experimentation 110
wire mother @echo.northsky.social · 8hgoing for smth like what if caves of qud/dwarf fortress/rimworld was on an unstructured mesh for a sphere/ring/disc world. but at this point more just a tool than a game (that'll hopefully come later) 010
wire mother @echo.northsky.social · 8hwith procedural generation stuff (like wave function collapse) 110
wire mother @echo.northsky.social · 8hat this point, it's basically a geospatial library for fictional (gamedev) data 130
wire mother @echo.northsky.social · 8hbut gray curves aren't continuous (same with morton). the hilbert conversion stitches it together to make it continuous. which makes KNN search much more efficient with guaranteed locality 010
wire mother @echo.northsky.social · 8hand gray curves (gray code + morton) are a simpler intermediary to hilbert curves. hilbert just adds local transforms to convert from gray index to hilbert. inverse gray coding is more expensive than forward tho. fwd is just y = x^(x>>1) 210
wire mother @echo.northsky.social · 9hi drew from this code a bunch, but added some SIMD optimisation and wider bit capacity. but the morton-hilbert transform is not bit width dependent 110
wire mother @echo.northsky.social · 9hif ur interested, z-curve/morton (bit interleaving), which can then be bi-mapped to hilbert curve with LUTs. you can see some code here (links to explanatory articles in the readme) github.com/rawrunprotec...github.comGitHub - rawrunprotected/hilbert_curves: Fast Hilbert CurvesFast Hilbert Curves. Contribute to rawrunprotected/hilbert_curves development by creating an account on GitHub. 110
wire mother @echo.northsky.social · 9hmust remember that i never actually make anything without a defined objective. and "but i wanna play with it" is not one 010
wire mother @echo.northsky.social · 9hno echo, you do not need to make a lisp. it is not relevant to anything you're working on no matter how much you want to play with NaN boxing 100
wire mother @echo.northsky.social · 9hI've been reading too much abstract algebra theory coz it took me a minute to remember cardinality isn't just about infinities 010
wire mother @echo.northsky.social · 9hwell i guess it's "indexing" then? its nD to 1D in any case (space filling curve) 210
wire mother @echo.northsky.social · 9hand with 3d points it's like, oh o got a whole extra 32 bits i could use there. looking to see if using that for vector magnitude improves locality for volumetric data. at the least i can use it for layer/height sorting (not interleaved with xyz) but multi layer locality would be cool 000
wire mother @echo.northsky.social · 9hdoes it even count as a hashing algorithm if it's bijective? 210
wire mother @echo.northsky.social · 9hbut probably good to minimise indirections. tho with the hashing, it should be fairly local memory access regardless, so cache misses shouldn't be a major issue 100
wire mother @echo.northsky.social · 9hre: working better with triangulations, that's mostly just about having consistent 3-edge vertices instead of variable, which allows for some nice optimisations. but not particularly necessary with half/quad/winged edges 100
wire mother @echo.northsky.social · 9htho, i would like to investigate mixed area/radius cells with WFC. could be interesting 100
wire mother @echo.northsky.social · 9hspatial hashing + quad edge rhombille should be plenty performant for WFC. don't see why the need for hex grid rhombilles. any delaunay/voronoi grid should work with area/radius constraints. and those constraints are implicit with poisson point gen, so doesn't require much 100
wire mother @echo.northsky.social · 9hwhy are edge duals not more of a thing? quad edge meshes are basically canonical mesh structures already, implicit quads already there. 100
wire mother @echo.northsky.social · 9hlooking forward to confirming rhombille/quad tile generation from arbitrary polygon meshes by using edge vertices + adjacent face centroids. can't see any reason why not, and will mean unstructured grids can produce quad tiles. should work especially well for triangulations, but any poly works 100
wire mother @echo.northsky.social · 9hneed to test but should be good for approximate hamilton paths too (hilbert indexing), which is very useful for fast point/face loop detection 110
wire mother @echo.northsky.social · 9hspatial hashing code completed, with some simd optimisation to boot 🙂↕️ 110
wire mother @echo.northsky.social · 10hsometimes i forget that i am not just a chemical machine but a person, and so i do actually need art and pleasure to live. 020
wire mother @echo.northsky.social · 20hgodzilla doesn't need to rely on a respiratory transport system because he has a reactor inside him (he needs to expel a lot of excess heat tho), but what about the other kaiju? do they all use some fission based anaerobic respiration? perhaps mothra's wings are giant tracheole networks? 000
wire mother @echo.northsky.social · 21hhoney i shrunk the kids. they died of nitrogen narcosis. honey i blew up the kids. they died of asphyxiation 000
wire mother @echo.northsky.social · 21hresident evil? oh is that about the darkness at the heart of suburban inhabitants qua lynch? 000
wire mother @echo.northsky.social · 21hif you found yourself in some kind of honey i shrunk the kids or downsizing situation, you'd probably just be high as fuck all the time from nitrogen narcosis 100
wire mother @echo.northsky.social · 21htheir respiratory capacity would be insufficient due to the square-cube law 000
wire mother @echo.northsky.social · 21hthey would no longer be well adapted to their evolutionary niche and struggle to survive 🙂↕️ 110
Reposted by wire mother🌏🐶🦕🥳 Vegetable Gremlin ⍼👻🥦💥🏠♫ @vegetablegremlin.bsky.social · 05/10/2026 81144347
wire mother @echo.northsky.social · 05/10/2026the non-spherical stuff will actually be easier to implement tbh. no weird patch boundaries to deal with 000
wire mother @echo.northsky.social · 05/10/2026ok, so my game project is sorta becoming a geospatial data tool for fictional worlds (with options for different topology like discs and rings) maybe it'll just be a tool 100
wire mother @echo.northsky.social · 04/10/2026they should invent a kind of mathematics that isn't fun or interesting 110
wire mother @echo.northsky.social · 03/10/2026what if we just punched a hole straight through the earth along the polar axis, just so it was easier to project onto a grid 000
wire mother @echo.northsky.social · 03/10/2026feel like i understand a bit better how floats relate to integral domain stuff now. one day i might even figure out automatic differentiation 000
wire mother @echo.northsky.social · 03/10/2026hard to learn certain stuff when you don't know the words to look for 100