Freya Holmér
@freya.bsky.social
🎮 indie tech artist
🏗️ I made Shader Forge & Shapes
🌐 working on https://half-edge.xyz
🔥 shader sorceress
📏 math dork
🎥 rare YouTuber/streamer
📡 ex-founder of @NeatCorp
my kids:
🥪 @toast.acegikmo.com
🥗 @salad.acegikmo.com
🐈⬛ @thor.acegikmo.com
🏗️ I made Shader Forge & Shapes
🌐 working on https://half-edge.xyz
🔥 shader sorceress
📏 math dork
🎥 rare YouTuber/streamer
📡 ex-founder of @NeatCorp
my kids:
🥪 @toast.acegikmo.com
🥗 @salad.acegikmo.com
🐈⬛ @thor.acegikmo.com
this morning, toast jumped the gate and ran directly towards gumpy. a fight broke out, and as I tried to break it up, even salad, who's been friends with gumpy, aggressively joined to fight her for the first time :(
everyone is (mostly) physically okay, but this is the first time there's been blood
everyone is (mostly) physically okay, but this is the first time there's been blood
February 5, 2026 at 2:33 AM
this morning, toast jumped the gate and ran directly towards gumpy. a fight broke out, and as I tried to break it up, even salad, who's been friends with gumpy, aggressively joined to fight her for the first time :(
everyone is (mostly) physically okay, but this is the first time there's been blood
everyone is (mostly) physically okay, but this is the first time there's been blood
I was trying to wrap my head around what polylines with bevel joins would even look like in 3D and uuuuuh I guess this makes sense
February 4, 2026 at 2:40 AM
I was trying to wrap my head around what polylines with bevel joins would even look like in 3D and uuuuuh I guess this makes sense
when naming hits just right
![C# code that reads:
[SerializeField] T[] dataDisk;
NativeList<T> dataCpu;
GraphicsBuffer dataGpu;](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:5nq3pybl4nnoxfp3ovjy2lh7/bafkreicpr7ive4727iot2omqqnbtn2oakqsxjnckjhqebcijmb2rr4l2qu@jpeg)
February 3, 2026 at 2:26 PM
when naming hits just right
it was time to overhaul the polyline rendering in Shapes!
it's now using primitive quads with structured buffers, which is gonna allow me to fix basically *everything* that was buggy/weird about the old mesh-based polylines
it's now using primitive quads with structured buffers, which is gonna allow me to fix basically *everything* that was buggy/weird about the old mesh-based polylines
February 2, 2026 at 11:36 PM
it was time to overhaul the polyline rendering in Shapes!
it's now using primitive quads with structured buffers, which is gonna allow me to fix basically *everything* that was buggy/weird about the old mesh-based polylines
it's now using primitive quads with structured buffers, which is gonna allow me to fix basically *everything* that was buggy/weird about the old mesh-based polylines
some people probably don't like it when i call my cats my children but what else am I supposed to do when there's this little lady wandering around my home asking for additional dinner
February 1, 2026 at 3:50 AM
some people probably don't like it when i call my cats my children but what else am I supposed to do when there's this little lady wandering around my home asking for additional dinner
surreal to finally meet this lil gal after having spent months looking for her
equally weird to break into tears at a stranger's house but like 😭 it's her, Siri is home!!
equally weird to break into tears at a stranger's house but like 😭 it's her, Siri is home!!
January 30, 2026 at 2:44 AM
surreal to finally meet this lil gal after having spent months looking for her
equally weird to break into tears at a stranger's house but like 😭 it's her, Siri is home!!
equally weird to break into tears at a stranger's house but like 😭 it's her, Siri is home!!
another issue with higher degrees is that they start to become unwieldy to solve by hand, which is why I've been looking into sympy/symbolic math library stuff!
a lot of these equations have like, factorial growth in size, so they're kinda nontrivial to compute
the uniform variant is clean tho!
a lot of these equations have like, factorial growth in size, so they're kinda nontrivial to compute
the uniform variant is clean tho!
January 28, 2026 at 1:14 AM
another issue with higher degrees is that they start to become unwieldy to solve by hand, which is why I've been looking into sympy/symbolic math library stuff!
a lot of these equations have like, factorial growth in size, so they're kinda nontrivial to compute
the uniform variant is clean tho!
a lot of these equations have like, factorial growth in size, so they're kinda nontrivial to compute
the uniform variant is clean tho!
an interpolating C² continuous quintic spline
January 28, 2026 at 12:59 AM
an interpolating C² continuous quintic spline
I just want my polynomial in ascending order and with the coefficient on the right but this is apparently a nightmare
and i can hear you all say "oh just use latex(expr, order='rev-lex')" but you see-
and i can hear you all say "oh just use latex(expr, order='rev-lex')" but you see-
January 27, 2026 at 3:46 PM
I just want my polynomial in ascending order and with the coefficient on the right but this is apparently a nightmare
and i can hear you all say "oh just use latex(expr, order='rev-lex')" but you see-
and i can hear you all say "oh just use latex(expr, order='rev-lex')" but you see-
im losing my mind i stg it's impossible to format equations to look nice in sympy/ipynb
January 27, 2026 at 3:42 PM
im losing my mind i stg it's impossible to format equations to look nice in sympy/ipynb
asdadfsafg sympy pls i just wanted to move the u^3 to the left side this is an EVIL interpretation of unevaluated expressions 😭
January 27, 2026 at 3:12 PM
asdadfsafg sympy pls i just wanted to move the u^3 to the left side this is an EVIL interpretation of unevaluated expressions 😭
i found some snow in case you need it
January 27, 2026 at 12:28 PM
i found some snow in case you need it
I was wondering why python was taking an hour to process this equation but I suppose 116,699 characters is on the long end
January 27, 2026 at 2:11 AM
I was wondering why python was taking an hour to process this equation but I suppose 116,699 characters is on the long end
siri is back home ❤️
January 23, 2026 at 11:32 PM
siri is back home ❤️
salad doesn't usually snuggle up close though (only when he wants to bother me at the computer)
he's a little different
he's a little different
January 22, 2026 at 2:11 AM
salad doesn't usually snuggle up close though (only when he wants to bother me at the computer)
he's a little different
he's a little different
this tweet was written during tonight's trickle
by the time I hit send, my arm had gotten fully surrounded by cats
by the time I hit send, my arm had gotten fully surrounded by cats
January 22, 2026 at 1:56 AM
this tweet was written during tonight's trickle
by the time I hit send, my arm had gotten fully surrounded by cats
by the time I hit send, my arm had gotten fully surrounded by cats
every night, all my boys eventually trickle in to the bedroom to nap together with me, it's adorable
cats are so deeply social, they just don't show it as acutely as dogs. but if you give them time and pay attention, you start to see just how much they want to spend that time with you ❤️
cats are so deeply social, they just don't show it as acutely as dogs. but if you give them time and pay attention, you start to see just how much they want to spend that time with you ❤️
January 22, 2026 at 1:49 AM
every night, all my boys eventually trickle in to the bedroom to nap together with me, it's adorable
cats are so deeply social, they just don't show it as acutely as dogs. but if you give them time and pay attention, you start to see just how much they want to spend that time with you ❤️
cats are so deeply social, they just don't show it as acutely as dogs. but if you give them time and pay attention, you start to see just how much they want to spend that time with you ❤️
gonna do this for now, I think it's generally fast!
1. assume a linear uniform Y value distribution
2. find gy, the y index of yt
3. use index gy as an initial guess gx for the interval we want to find
4. linear search in the direction of the target value
in my example, it's just one iteration
1. assume a linear uniform Y value distribution
2. find gy, the y index of yt
3. use index gy as an initial guess gx for the interval we want to find
4. linear search in the direction of the target value
in my example, it's just one iteration
January 20, 2026 at 6:14 PM
gonna do this for now, I think it's generally fast!
1. assume a linear uniform Y value distribution
2. find gy, the y index of yt
3. use index gy as an initial guess gx for the interval we want to find
4. linear search in the direction of the target value
in my example, it's just one iteration
1. assume a linear uniform Y value distribution
2. find gy, the y index of yt
3. use index gy as an initial guess gx for the interval we want to find
4. linear search in the direction of the target value
in my example, it's just one iteration
if I have an array of y values, is there an O(1) way to find the corresponding x value xt, given a y value yt?
it's basically a root solve, but it's discretized and linear in a way where I feel like there has to be a better shortcut, but maybe there isn't?
it's basically a root solve, but it's discretized and linear in a way where I feel like there has to be a better shortcut, but maybe there isn't?
January 19, 2026 at 10:50 PM
if I have an array of y values, is there an O(1) way to find the corresponding x value xt, given a y value yt?
it's basically a root solve, but it's discretized and linear in a way where I feel like there has to be a better shortcut, but maybe there isn't?
it's basically a root solve, but it's discretized and linear in a way where I feel like there has to be a better shortcut, but maybe there isn't?
US politics is so stupid right now it's insane
trump doesn't want greenland for national security reasons
he wants greenland because he wants greenland
trump doesn't want greenland for national security reasons
he wants greenland because he wants greenland
January 18, 2026 at 4:44 PM
US politics is so stupid right now it's insane
trump doesn't want greenland for national security reasons
he wants greenland because he wants greenland
trump doesn't want greenland for national security reasons
he wants greenland because he wants greenland
speedrunning is weird bc you can start doing things like casually play through Silksong over breakfast
January 17, 2026 at 11:06 AM
speedrunning is weird bc you can start doing things like casually play through Silksong over breakfast