TheSeanLavery retweeted
Replying to @CaptainCH76 @justalexoki
its hard to automate intelligence without AI i guess
TheSeanLavery retweeted
My Linux journey started in 1996 with the LinuxPPC project. At the time it was a fork of the kernel that was able to boot the Power Macintosh 8500 I had into a full-blown Linux environment.
The journey I embarked upon to get up and running (aka downloading 500mb of rpms over a 33.6 modem) started a love affair with Linux and open source software that changed the trajectory of my life.
Flash forward 30 years and I have an opportunity to help make Linux on the Macintosh something special
I'm humbled and deeply grateful for the opportunity to work with the @OmarchyMac team and the broader community to help make Linux on the Macintosh more incredible than its ever been before.
Thank you for extending the invite, @dhh and thanks for sparking a movement that I firmly believe will restore open source and Linux to its roots
Let's fix everything 💪.
The Omarchy Doctrine
Unite the nerds
Hold the line
Have some fun
Beauty is truth
Heritage is duty
Command is service
Welcome the agents
Perfect the computer
Own the machine
You’re somebody now
omarchy.org/doctrine/
TheSeanLavery retweeted
Touch ID on Linux, on a MacBook Pro with NO Apple firmware left. The installer had wiped the T1's EFI data, no macOS anywhere, everyone said reinstall macOS.
Got the T1 to regenerate its own data from Linux. It booted. @0xBOYD's t1bridge took it. Fingerprint works. @OmarchyLinux
TheSeanLavery retweeted
Replying to @davesoderberg
on other videos they’ve got, it’s shown opened like a book and the effect it’s preserved. i’m guessing usage of gyroscope to at least track the front
TheSeanLavery retweeted
Replying to @rouwnee
Mario 64 6 seconds after game starts:
youtube.com/watch?v=hk7egira…
TheSeanLavery retweeted
the future of design is programmatic:
different from generative AI
different from diffusion image generation
instead of asking a model for finished pixels, you describe the rules of a visual system, then compile, test, revise, and reuse it
a thread on my process:
TheSeanLavery retweeted
ESP32 Hacking with @CyborgLavery
A portable codex that I can talk to and it talks back to me
TheSeanLavery retweeted
To supplement assets that are harder to procedurally generate like props, buildings, etc, there will be a few integrations available such as @polyhaven (CC0 license) and @ambientCG (CC-BY license). What this will do is giving a way for us to bootstrap available assets.
As you can see in this video, the defined shader automatically applies in the preview so you can see what it looks like vs the original (default shader in this case is the @callmesenseieng one). I would say our shader does make some of these objects look even better than the original, but I'm also a bit biased...LOL
Coupling this with MCP, you all the sudden have what people have been using with Blender x MCP. This idea essentially came out from trying to figure out why some people were able to achieve a much better result with Blender x MCP than others. I believe with the right setup, you can now also do crazy things quite quickly directly in @threejs.
Will do more testing and set things up so it's as easy as possible to use. The UI is still a bit all over the place for the labs and I will probably look to unified them a few versions in. A bit busy with open beta launch for Call Me Sensei app this week so things are a bit delayed, but it also gave me a bit more time to explore and learn about how others do things.
Continuing with the series of tools for game dev on @threejs. We are now at our next stop which is procedural texture generation lab. There is also an AI-assisted tool (BYOK) where you can describe what you want like "old leather jacket, sun bleached with scuffed seams" and it will figure out all the params for you. You can also upload and image and it will use some crazy math to generate a texture file for you. Future enhancements will also allow you to upload screenshots of a full environment and it can figure out how many textures you will need and generate them for you in one go.
Part of the open source library (v0.2.0) is now being tested in the wild (including by @callmesenseieng team) and it's quite clear that additional documentation and examples are required for LLM to effectively use it. We are on target to release a more optimized version late this week.
























