@Volorf I like this a lot. Using a tiny ESP32 device as a physical controller for a VisionOS game makes the interaction feel much more tangible than keeping everything inside the headset.
@computeontology The webcam-as-receiver part is really neat. I’m curious how you handle synchronization between the LED channels and the camera frames.
@BhosalePratim The physical button is what makes this work for me. No app to open, no screen to navigate, just press and talk. That kind of zero-friction interaction is where dedicated little devices really shine.
@erfmufn Cool, 20 to 30 FPS over SPI is honestly pretty good for that little display. Chasing 40 sounds like one of those optimization rabbit holes that could be hard to resist.
@techexplain1 The appeal is that you can dedicate one to a single physical job without overthinking the cost. Read a sensor, drive a display or motor, connect it to WiFi, then leave it running.
@tom_doerr Seeing an ESP32 actually keep something in the air never stops being impressive. Flight control exposes every little timing, sensor and tuning problem at once.
🚀 JARU IDE is live on Product Hunt today!
I used to build my ESP32 projects in C/C++. It works, but for quick experiments I was spending more time fighting toolchains and libraries than writing actual code. So I started a small language with the libraries I use most already built in.
That "small project" turned into a full system:
🦎 JARU, a language made for microcontrollers
⚙️ A virtual machine that runs on the ESP32 and on Windows
🔧 A compiler targeting that VM
🐞 Debugger with breakpoints and variable watch
💡 GPIO simulator, to test LEDs, actuators and panels with no board attached
🎨 Image, sprite and tilemap editors
🎵 A built-in chiptune music tracker
📡 MQTT support
⚡ One-click flashing to the board
The idea: write your code, run it and step through it on your PC first, and only flash the board once it actually works. No heavy toolchain, no long setup.
It turns out to be a really fast way to prototype MQTT integrations, REST APIs, sensor tests and small retro-style games. I have been using it for my own demos, including a recent ray tracing experiment.
Happy to go deep on any part of it: the language design, the VM, the compiler or the simulator. And I am taking requests, so which boards, displays or peripherals should I support next?
Upvotes and comments over there really help 🙌
👉 https://t.co/kJVnSOUA5G
@mlesplmer@steveruizok Taking a generative piece from a Raspberry Pi to an ESP32 should be an interesting shift. Curious to see how the idea evolves once you start building the new version.
@daniel_nguyenx@steveruizok Building the personality before adding the voice feels like the right order. If the expressions already make your son react to it, the little robot is doing its job.
@madietlx Nice choice for moving from software into hardware. The Nano form factor feels familiar, but having an ESP32-S3 underneath gives you a lot of room to explore once you get past the first few experiments.
@tom_doerr I still find it fascinating that WiFi can be treated as a sensor instead of just a connection. Combining CSI with vision for real-time 3D pose estimation is a very clever use of data that normally gets ignored.
@Alacritic_Super I like the balance here. A modular multimedia pipeline is useful, but keeping the runtime overhead around 7 KB is what makes it genuinely interesting on an ESP32.
@hot_town@steveruizok That is a seriously ambitious first ESP32-S3 build. Starting with audio, a codec and e-paper instead of the usual blink demo is one way to learn the platform quickly.
@DynamicWebPaige@googlegemma@github The clever part is not just fitting 28.9M parameters into an ESP32-S3, but avoiding the need to keep them all in fast memory. Pulling only the embedding rows needed for each token is a brilliant workaround.