Hand tracking based full body #VR #fitness with jumping, pushups, crunches and more
| Website | https://xrworkout.io |
| Discord | https://chat.vrworkout.at |
| AppLab | https://app.xrworkout.io |
| https://twitter.com/XRWorkout |
I probably should do a post mortem of our menu rebuilding over the last two months, but that's for another day.
First we want to share the early signals we are getting.
Those are two of our weekly workouts. The top one was our most successful weekly workout before the menu refresh.
And below is our first weekly workout after the new menu was released.
The engagement of the users is "off the charts" with the latest weekly workout doing in 6 days well above what the other one did in 3 months.
Experimenting with local voice models running directly on the Samsung Galaxy XR.
The results are really promising. I did not get any stutter in the tracking which makes me hopeful that we have enough performance headroom left.
The goal is to replace the Meta Voice SDK with something running completely locally on device, this will also greatly advance our Workout Buddy when player can tell their virtual personal trainer what workout they want it to build.
The Flax Engine just released Flax 1.11 with several core improvements to the shocking capable game engine with C++ and C# scripting.
You can think of Flax like an Indie Unreal Engine, they are very similar in many key ways.
In the future I plan to support standard USB HDMI capture cards as well, that will allow it to be used with other SBCs that don't have a HDMI-in port like a Raspberry Pi
It's an early proof of concept and the performance isn't too good yet, but the code is available here: https://github.com/mgschwan/viture_virtual_display
I built a virtual HDMI display with the VITURE Pro XR glasses through the HDMI-in on an OrangePI 5 Plus
Now I can connect my headless Mac Mini via HDMI to the OrangePi 5 Plus, which maps the HDMI frame onto a virtual screen based on the head rotation that is read from the Viture glasses
One little snag was that I needed to reverse engineer the communication with the glasses to get the IMU data because the official Viture SDK is only for x86 but the OrangePi is ARM based
I've completely remade City Kit - Suburban! This pack includes 40 house models, also includes color variations and it's CC0 public domain so completely free to use in your projects ✨
Power consumption of our XO gym companion prototype.
We are aiming for a battery life of 4 hours
Majorly impressed with DeepSeek R1
The phrase was generated with r1:14B on my local old 3070 laptop GPU with 6GB VRAM
Prompt was: Complete the sentence: To be or not to be and put it into a fitness context
It kept iterating on several ideas automatically and landed on that quote
Image: Flux 1.1 Pro
The vOICe vision BCI 🧠🇪🇺
GameFromScratch
Kenney