Devlog III

Hi everyone, it’s been a while. I really wanted to work on this (and other stuff too), but going back to school has been rough (IB is srsly no joke) so pretty much all my other projects have been on hold. I worked for a bit, but didn’t do as much as I would’ve liked. Honestly, right now this is kind of half-baked, but with Stardance ending I have no choice but to ship what I’ve got. Who knows, maybe I’ll work on it more in another program, but for now I’m putting everything on the backburner and focusing on my studies. If anyone else wants to take this project further plz do, cuz I would love to see a fully finished version of this idea :)

Devlog

Ok so I kind of locked in and one-shot this over the weekend, but here’s an overview of everything that I did. Basically I started with the schematic, which was 80% easy and 20% crazy rabbit-hole cuz I decided I really want the possibility to make this battery powered and to monitor the battery level. Afterwards I moved on to the PCB and routing, which didn’t take too long but I ended up spending wayyy too long on the silkscreen and finding a nice font and all. Anyways, I’m really happy with how it turned out, so it’s worth the 4-5 ish hours I put into it :)

CAD was awful at first. I was using FreeCAD, which is an awful experience imo (0 stars, would not recommend). Thankfully @EMT (formerly babydriver) saved me and showed me the joys of Onshape. I managed to get like 90% of the CAD done in about an hour and a half, which was awesome. I forget to record it with Lapse though 😭 There’s not really much more to say about the CAD, except maybe that I spent wayyy too long trying to make it perfectly symmetrical for the display so that it looks nice.

Finally for the firmware I used C++ with Platform IO in VS Code (yes I started Hackatime this time). It’s currently a first draft, as I need to have the parts to do some testing before I can continue. However as of now it supports going into deep sleep (safely hopefully), being woken by a keypress, and drawing a date, time, and battery percentage to the TFT display. I’m planning to add alarms, time sync over Wi-Fi, alarms, 12H-24H time toggling, and a little config over an AP (probably just as a website). Again though, I’m really stuck right now cuz of the charlieplexing (kind of shot myself in the foot lol). I understand the theory and all, but I just need to see it before I can work with it.

Anyways, that’s all for now. If anyone has any feedback that would be appreciated. If not, then thanks for reading this far and see y’all in the next one :)

Devlog II

Ok so I locked in for the past few days and got as much done as possible to the point of being ready to ship. I’ve made a case design using Blender (cuz apparently STL files are also fine), and struggled through the writing of QMK firmware. I tried reading the docs but cuz I used charlieplexing to wire my keys the docs were kinda useless so I had to fallback to AI, which honestly wasn’t much better.

So yeah anyways a few hours later and I now have a v1 of my case and a v1 of my firmware. My firmware is currently missing support for the OLED and LEDs, but from what I can tell those should be fairly easy and I want to test my keys and rotary encoders by themselves before I go making the scripts any messier than they already need to be. Same goes for the case. What I have now is a good start (especially cuz I haven’t done much CAD before), but I want to make it better. I’ll do a v2 design once I receive parts and have a clearer idea of how I want everything to fit together :)

For now I’m gonna ship and hope it all goes well (though it probably won’t lol)

Devlog I

Took me a while but I managed to get the schematic and PCB design done. I ended up redoing the schematic twice cuz I kept trying to expand it to make more stuff fit with the limited number of logic pins. The PCB design went fairly smoothly until I finished my first run and realized that it was too wide… so uh yeah I ended up redoing that too lol but it’s all good now. Gonna try to do the case tmrw and hopefully the firmware and everything will be done by the end of the week :)

Devlog II

So after managing to get MediaPipe working fairly quickly yesterday I decided to try to start working on the 3D rendering aspect of the project. I’m using three.js to do this in the browser with JS, but it has a bit of a learning curve so I’m struggling a bit. Rendering a simple cube is fine but getting an STL model I got from printables on the screen took me a lot longer than I anticipated

Anyways, the next steps to take will be to get the model rotating and resizing based on the detected head size and orientation. Turns out that MediaPipe gives me the matrixes, and that three.js comes with built-in functions for this (sweet), except this means that I need to merge my two JS scripts into one thingy, which is not a fun process. I started today, but I need some recovery time before I finish

After that I’ll probably need to do some work on the models, adding colors and all that. I’m planning to use Blender, but if anyone has any better ideas/suggestions plz do let me know. Also idk if it’s possible to add Hackatime to Blender, so that’s smthn imma have to figure out too… Basically I’m happy with how it’s going, but there’s still a ways to go

If you wanna check out the half-baked stuff I’ve got so far, I so far have a head-tracking demo and a Iron Man helmet rendering demo. As always, I’m happy to hear if y’all have any ideas on how I can improve :)

Devlog I

So I’m building suit-up, which is a website that should let you overlay Iron Man onto you. Started this cuz I keep seeing ppl building iron man and spiderman stuff on my insta, and cuz I’m not skilled enough to do that I decided this would be the next best thing.

I started with Python, except then realized that building a tool that runs an AI model and renders 3D models in parallel is gonna suck and kill my laptop. After some research I found that I could either do this in the web (as there are libraries that let me do this efficiently), or I could spend hours struggling with C++… the choice was pretty easy honestly

Anyways, I’ve now gotten a cameo working with an overlay of the landmarks that are detected by the mediapipe model I’m using. Had to use Gemini to understand the docs for the model as they’re cryptic as heck and incredibly overcomplicated. Next up I’m gonna see how I can render a 3D model, and then imma try to implement that as an overlay. Got a long way to go, but I think it’s gonna be cool

Btw, there is actually a cameo but I made it transparent for the screenshot cuz it’s still too early for a face reveal lol

You can try my really rough demo here if you’re interested. All feedback is appreciated, but plz keep in mind this is still in early dev stages

Devlog

So over the past three days I’ve been building beatLock. Basically I resuscitated my old 2013 macbook pro with Arch Linux, but was kinda upset it didn’t have touch ID like my current one does. Long story short: I decided to try to build a form of biometric authentication based on typing patterns, with the idea that this makes passwords safer.

I started with a web demo in HTML/CSS/JS, mostly cuz I know those already and I rlly just wanted to figure out how this thing was gonna work without dealing with syntax errors every five mins. It only took me abt an hour to get this web demo working which was nice. Unfortunately it’s not too polished (sorry abt that), cuz the focus was getting it working on Linux.

Then I started coding in C, which in and of itself ain’t awful cuz I work with C++ regularly. Thing is, you can’t just take input normally, so I sent a couple hours going down a rabbithole of getting keyboard events directly from /dev/input/ (the low-level kernel events basically).

Ironically, the part that screwed me over the most was the part I hoped would be rlly easy. Turns out that writing protected files from a hidden root process brings with it a lot of security considerations, so I had to get AI to help me build smthn that was acc secure instead of the half-baked security nightmare fuel I built. I spent wayy too long on that and compiler errors, but I digress…

Anyways, after going through a couple of iterations of integration hell with the PAM library, I got it working on Linux. Building the Makefile and installation scripts wasn’t too difficult (especially compared to C cough cough). Add a README and a MIT License, and we’re good to go

Basically, the result is a functional biometric typing pattern authenticator that can be installed on Linux (instructions in the README). Currently I’ve got it set up to still fallback to the default PAM (just in case yk), but it is possible to set it as the sole PAM if you want to (see the experimental section of the README). No idea if people will actually wanna use it, but I had fun building it. Hope y’all like it too :)

P.S. For all the Mac users, I found out you can use touch ID to authenticate sudo on MacOS. Here’s the guide if you wanna try. Works like a charm for me