Jonathan Merritt
@lancelet.bsky.social
AI, deep learning, numerical, functional programmer.
Looking for a job!
Uilleann piper, highland piper, extreme Gaelic music enthusiast.
https://jmerritt.blog
Looking for a job!
Uilleann piper, highland piper, extreme Gaelic music enthusiast.
https://jmerritt.blog
No Strava, nooooooooo!
(But that’s a nice way to do an outage message.)
(But that’s a nice way to do an outage message.)
April 23, 2025 at 8:31 AM
No Strava, nooooooooo!
(But that’s a nice way to do an outage message.)
(But that’s a nice way to do an outage message.)
There’s no extra context here. I just asked for which conveys the lower likelihood. I would certainly say that “improbable” is not the same as “unlikely”, but I’m not American.
March 17, 2025 at 3:40 AM
There’s no extra context here. I just asked for which conveys the lower likelihood. I would certainly say that “improbable” is not the same as “unlikely”, but I’m not American.
“There’s a WHAT?! In AMERICA?!!!” 😁
March 11, 2025 at 6:32 AM
“There’s a WHAT?! In AMERICA?!!!” 😁
In Nike Run Club, there’s a looong way between Black and Volt levels!
February 28, 2025 at 8:33 AM
In Nike Run Club, there’s a looong way between Black and Volt levels!
WinderBot’s very first winding. I only got this far before I accidentally broke the copper wire.
Still, I learned a few things for the next try, and I’ll be able to use this one to confirm some magnetic field strength estimates.
Still, I learned a few things for the next try, and I’ll be able to use this one to confirm some magnetic field strength estimates.
February 15, 2025 at 5:54 AM
WinderBot’s very first winding. I only got this far before I accidentally broke the copper wire.
Still, I learned a few things for the next try, and I’ll be able to use this one to confirm some magnetic field strength estimates.
Still, I learned a few things for the next try, and I’ll be able to use this one to confirm some magnetic field strength estimates.
The 6 month view is not great either. Could easily wipe out those gainz in a month if the negative sentiment continues.
They’re losing sales worldwide, and those high prices were predicated on the fictitious continued growth: www.abc.net.au/news/2025-02...
They’re losing sales worldwide, and those high prices were predicated on the fictitious continued growth: www.abc.net.au/news/2025-02...
February 12, 2025 at 5:13 AM
The 6 month view is not great either. Could easily wipe out those gainz in a month if the negative sentiment continues.
They’re losing sales worldwide, and those high prices were predicated on the fictitious continued growth: www.abc.net.au/news/2025-02...
They’re losing sales worldwide, and those high prices were predicated on the fictitious continued growth: www.abc.net.au/news/2025-02...
Make sure he has his adderall.
February 4, 2025 at 11:22 AM
Make sure he has his adderall.
CNC pong. The most exciting game my coil winding robot will ever play.
February 4, 2025 at 6:36 AM
CNC pong. The most exciting game my coil winding robot will ever play.
Custom shiny butts for my steppers!
February 1, 2025 at 11:20 AM
Custom shiny butts for my steppers!
WinderBot progress: the linear motor mount is finished.
January 31, 2025 at 1:24 AM
WinderBot progress: the linear motor mount is finished.
I’m currently building the world’s most “silly” CNC machine. It’s going to wind electromagnets.
Don’t get me wrong… my estimate is that it will save ~100h of crushing manual labor; and do a better job. But it’s still “silly”. Who builds a CNC machine at home for this? 🤷♂️
Don’t get me wrong… my estimate is that it will save ~100h of crushing manual labor; and do a better job. But it’s still “silly”. Who builds a CNC machine at home for this? 🤷♂️
January 29, 2025 at 12:24 PM
I’m currently building the world’s most “silly” CNC machine. It’s going to wind electromagnets.
Don’t get me wrong… my estimate is that it will save ~100h of crushing manual labor; and do a better job. But it’s still “silly”. Who builds a CNC machine at home for this? 🤷♂️
Don’t get me wrong… my estimate is that it will save ~100h of crushing manual labor; and do a better job. But it’s still “silly”. Who builds a CNC machine at home for this? 🤷♂️
My WGPU line renderer can now handle different end cap styles, and drawing lines in order (bottom to top).
It also has a place on GitHub now, although it's very much WIP still: github.com/lancelet/bea...
It also has a place on GitHub now, although it's very much WIP still: github.com/lancelet/bea...
January 14, 2025 at 7:53 AM
My WGPU line renderer can now handle different end cap styles, and drawing lines in order (bottom to top).
It also has a place on GitHub now, although it's very much WIP still: github.com/lancelet/bea...
It also has a place on GitHub now, although it's very much WIP still: github.com/lancelet/bea...
The glow is entirely fake (it's part of the width of the line). Here's the width of lines as the engine sees them.
I'm strongly considering writing my own asteroids clone with this. :-)
I'm strongly considering writing my own asteroids clone with this. :-)
January 10, 2025 at 12:55 PM
The glow is entirely fake (it's part of the width of the line). Here's the width of lines as the engine sees them.
I'm strongly considering writing my own asteroids clone with this. :-)
I'm strongly considering writing my own asteroids clone with this. :-)
I'm using a tiled scaffold approach, similar to PathFinder or Vello (but obviously much simpler).
There's a pre-processing step on the CPU to find occupied tiles, which are instanced. Then the fragment shader looks up the lines for the SDF.
(2/3)
There's a pre-processing step on the CPU to find occupied tiles, which are instanced. Then the fragment shader looks up the lines for the SDF.
(2/3)
January 10, 2025 at 12:54 PM
I'm using a tiled scaffold approach, similar to PathFinder or Vello (but obviously much simpler).
There's a pre-processing step on the CPU to find occupied tiles, which are instanced. Then the fragment shader looks up the lines for the SDF.
(2/3)
There's a pre-processing step on the CPU to find occupied tiles, which are instanced. Then the fragment shader looks up the lines for the SDF.
(2/3)
I've been exploring rendering lines in WebGPU / WGPU using signed distance fields.
Some initial results showing lines with a "glowing" effect.
(1/3)
Some initial results showing lines with a "glowing" effect.
(1/3)
January 10, 2025 at 12:52 PM
I've been exploring rendering lines in WebGPU / WGPU using signed distance fields.
Some initial results showing lines with a "glowing" effect.
(1/3)
Some initial results showing lines with a "glowing" effect.
(1/3)
Satellite-AIS correlation to track the global shadow fleet.
New blog post: jmerritt.blog/blog/2025010...
Some thoughts I had about tracking the extent of Automatic Identification System (AIS) manipulation using satellite images, and a couple of possible data sources I found.
New blog post: jmerritt.blog/blog/2025010...
Some thoughts I had about tracking the extent of Automatic Identification System (AIS) manipulation using satellite images, and a couple of possible data sources I found.
January 5, 2025 at 2:58 AM
Satellite-AIS correlation to track the global shadow fleet.
New blog post: jmerritt.blog/blog/2025010...
Some thoughts I had about tracking the extent of Automatic Identification System (AIS) manipulation using satellite images, and a couple of possible data sources I found.
New blog post: jmerritt.blog/blog/2025010...
Some thoughts I had about tracking the extent of Automatic Identification System (AIS) manipulation using satellite images, and a couple of possible data sources I found.
A neat part is, if you ask ChatGPT, it totally plays along! 😆
December 9, 2024 at 12:45 PM
A neat part is, if you ask ChatGPT, it totally plays along! 😆
Happy ShanksGiving everyone!
November 30, 2024 at 3:18 PM
Happy ShanksGiving everyone!
The Kwisatz Cabbage Patch.
November 19, 2024 at 6:26 AM
The Kwisatz Cabbage Patch.