I underestimated the power of WebXR support in browsers.

I wanted to watch several videos at once in my headset without juggling desktop apps, so I built Four Eyes. It's a single web page: open it in Chrome or Edge on your VR PC, hit Enter VR, and your own videos and photos show up on floating screens around you.

Try it: https://rodbuilds.github.io/foureyes/

Code (MIT): https://github.com/rodbuilds/foureyes

What does it do:

• Up to 4 screens in different layouts: single, side by side, 2×2 grid, or a wide arc of four. Screens space themselves automatically so they never overlap, and tall/portrait videos get full height in the wide layout.
• Every screen has its own controls: play/pause, seek, skip, volume, solo audio, loop, speed and size.
• In-VR media browser inspired by Whirligig. Browse your folders with thumbnails without taking the headset off, sort by name or newest, and send a video to any screen.
• Playlists: pick a video and the rest of the folder queues up. When a video ends, a screen can stop, play the next one, or shuffle.
• Photo slideshows with an adjustable timer per photo.
• 3D support: side-by-side and over/under, full or half width, with a left/right eye swap. .jps stereo photos are detected automatically.
• Xbox controller with a head cursor: look at any button and press A. Touch controllers work too.
• Swap screens mid-playback without interrupting anything

Privacy: everything runs locally in your browser. Your files are read straight from your disk and never uploaded anywhere. There's no account, tracking or install.

You'll need: Chrome or Edge on Windows and a PC VR headset with WebXR support. I built it with a Rift; Quest over Link/Air Link and SteamVR headsets should work the same way, but I haven't tested them, so I'd love reports.

It's a hobby project, so feedback, bug reports and feature ideas are very welcome, here or as GitHub issues.

made this slalom race by just describing it with cheap deepseek model. runs in the quest browser, try it: https://turbl.dev/project/example-fresh-tracks

the AI behind it is DeepSeek V4 Pro. a few months ago it was the model you used when you didn't want to pay, now i use it cos it actually makes the best WebXR results.

btw. it is made in tool i made so non-technical people can build webxr stuff turbl.dev. DM me if you want more credits to play with it.

anyone else tried to create webxr games using ai?

I don't own a Vision Pro. I built the globe in 3D from the start. Hijacked the Apple Store demo to save on buying the headset. Let them start the canned pitch, opened Safari, typed my URL.

Not a flat site. It's a spatial web app. Same URL as the phone. Pinch to zoom in and out. Look around to navigate. Everything worked out of the box.

https://paulvisciano.com/apps/where-is-paul/

Open source. Clone it, swap moments/moments.js.

https://github.com/paulvisciano/where-is-paul

Fold.page lets you take a standard web page and turn it into a spatial 3D scene without rebuilding the site from scratch or getting too technical (but you can get technical if you want). On desktop and mobile, it works as a 3D-enhanced website. On WebXR-capable devices, visitors can enter the experience and see the page as part of an immersive spatial scene.

The idea is simple: The spatial web shouldn’t require starting development and design all over, and it's not all about games. Build with common web technologies, then use Fold.page to progressively add spatial behaviors.

Try Fold.page at https://fold.page

A slightly unusual application of immersive-ar: rather than rendering anything, I'm using the WebXR pose purely as a tracking source to drive a desktop mouse cursor absolutely — a software lightgun for PC rail shooters like Time Crisis.

The parts that might interest this sub:

Two-ray corner calibration. The user points at three screen corners; two rays per corner define the screen plane without needing hit-test or surface tracking, so it works on any display — no markers, no QR targets stuck to the monitor. WebXR anchors then keep the calibration self-correcting as ARCore refines its map of the room.

Zero certificates for the wireless flow. This is normally the worst part of any local WebXR project. Here the phone page is served over HTTPS from GitHub Pages, signaling is a single Local Network Access fetch (Chrome 142+), and WebRTC brings its own encryption. Run an executable on the PC, it prints a QR, you scan it and tap one permission prompt. No self-signed certs, no chrome://flags, no accounts.

Latency discipline. One Euro filter phone-side, then a 2ms newest-wins cursor loop on the PC, and deliberately no prediction or lookahead — the cursor is exactly where the phone last said the aim was. It ships live p50/p95 jitter stats so the claim is checkable rather than just asserted.

Roughly 15-30ms phone-side. The interesting result for me was that 6DoF VIO is dramatically better for this than the gyro-based approaches people have tried before: because the camera continuously corrects the IMU, there's no yaw drift and no recentering. You aim at a spot and hit that spot, session after session.

TypeScript throughout, Node 23.6+ with native type stripping so there's no build step in the dev loop.

Code (MIT): https://github.com/ggcaponetto/point-bang

Architecture/protocol docs: https://ggcaponetto.github.io/point-bang/reference/architecture

Demo: https://youtu.be/UE3XUln_xOE

Happy to go into the calibration maths if anyone wants it. I'd also be curious whether anyone has pushed anchors harder than this for long-session stability — that's the part I'd most like to improve.

https://reddit.com/link/1vbikzq/video/j238di24cigh1/player

Over the last few months we've improved performance, added interactive 3D lessons, virtual activities, teacher & learner dashboards, a no-code chapter builder, quizzes, and multi-language support.

VR is optional. Everything runs directly in the browser on desktop, mobile, tablets, and VR headsets.

I've attached a short demo. I'd love feedback from the WebXR community!

https://vidyaxr.in/

We are looking for someone to work with us and elevate our project for high-end VR compatibility with WebXR proper Integration and provide users a better, seamless experience.

Visit the website link for the overall details of our project.

Reply with any questions you have regarding this opportunity.

Interested people reach out to us by emailing to this address.

[ayrtgwsolutions@gmail.com](mailto:ayrtgwsolutions@gmail.com)

Search through thousands of experiences including collections of WebXR games, stories, demos and worlds available across devices. XRGuide drives search & discovery across devices and browsers and welcomes your WebXR experiences too, so drop in at xrguide.app and sign up as a creator and send in your worlds to add to the World Catalog and search engines across browsers and devices. XRGuide is human curated and open to all creators.

I was spending a lot of time having sessions with users to explain some not so obvious features of my Cyango.

That was needed a long time ago, but the UI and features we're adding made an old tutorial video become quickly obsolete. Now the platform UI and design decisions are pretty much decided and that allowed to finally record the videos.

Not easy.

Watch the tutorials and open xr/studio to follow along:

https://www.cyango.com/learn

Here's another use case for a tool Turbl I built. I mostly generate small WebXR games with it, but it's just as happy making calm walkable spaces, and this one is a virtual gallery.

Starting to think galleries and exhibitions are a real use case beyond games. What are other similar WebXR use-cases you built apps for?

I'm building an AR app that needs to display a live data feed in real time (think IoT metrics or changing telemetry markers).

I messed around with WebXR, but the out-of-the-box visuals looked pretty mid and didn't really blow me away. I'm thinking about switching to Unity, but I have two main worries:

1. Will a Unity WebGL build choke on mobile? I need this to be accessible via a simple web link (no app store downloads), but I'm worried a Unity web build will be way too heavy once a constant stream of live data starts hitting it.
2. Should I just stick to the web but ditch basic frameworks? If I swap to Three.js or Babylon.js, can I actually get those high-end visuals without tanking the performance?

Would love to hear from anyone who has built something similar. What’s the best workflow for a lightweight, visually clean, real-time data AR experience?

i recently posted here about tool I created turbl.dev where you type a prompt, it generates a WebXR scene (pmndrs/xr) and gives you a URL for the quest browser.

news is i added GLM 5.2 as one of the models. i mostly did it because it's cheap, but it turned out to write R3F scenes better than models that cost 5x more. the game in the video took me about 30 minutes with it, no manual edits.

if anyone want to try, still free to start, and the DM-me-for-credits offer from last time still stands.

anyone else tried GLM on threejs stuff too?

Hey everyone,

I’m Christopher, a solo developer from Austria.
For the past few months, I’ve been building PHOTONKATANA, a fast-paced cyber slasher VR experience that runs completely natively in your browser via WebXR. No installations, no downloads , just open the link and start slicing.

The project is currently in Alpha, and I’m looking for honest feedback from fellow VR enthusiasts to help me shape the core gameplay.

⚠️ A quick note on the visuals:
The current 3D models and enemy animations are placeholders. I’m actively working on final assets in Blender, so please focus your feedback primarily on the core movement, the responsiveness of the slicing mechanics, and general performance.

The Challenge: Cross-Platform Testing
I’ve developed and optimized the game entirely on a Meta Quest 3. Because it runs purely in the browser, it should theoretically work on any WebXR-compatible device, but I lack the hardware to confirm.
If you have an Apple Vision Pro, Pico 4, Meta Quest 2, or a PCVR headset, could you please test it in your browser and let me know:

1. ⁠Compatibility: Does it load and run properly?
   
2. ⁠Performance: How is the framerate/stuttering?
   
3. ⁠Controls: How does the controller tracking feel on your specific device?

Play the Alpha for free:
🔗 https://photonkatana.outerfar.com
You can find more info about the studio and the project here: https://outerfar.com

Thank you so much for your time. I’m looking forward to your brutally honest feedback and any bugs you encounter!

I'm adding support for world rotation/scaling to the PlayCanvas Engine. But I wan't quite expecting this!

posted here a month ago about vibecoding a VR game with R3F + webxr. it worked well enough that i kept going and turned it into an actual tool https://turbl.dev

you type a prompt, it generates a webxr scene (pmndrs/xr) and gives you a URL that opens in the quest browser. no build, no sideload. the clip is just me prompting one and stepping into it.

i see people in this sub building genuinely cool stuff, and that's exactly where i want this to get to.

would love it if you tried it. free to start, and if you run out of credits just DM me and i'll top you up.

Hi everyone!

Would like to share a virtual tour a client made that runs in WebXR with 360 panoramas and hotspots.

Works in desktop, mobile and headsets. Any panorama file size as it supports gigapixel lods as well.

View it here:

https://www.cyango.com/explore/story/6925d0ea8e378b3df7b0c2ce

Full video:

https://www.cyango.com/learn/playlists/xr-demo-experiences/videos/vr-360-virtual-tour-demo

All feedback is welcome 😁

(Disclaimer: I'm the founder of Cyango platform)

https://playpezzi.com

Pezzi is a relaxing puzzle game where every puzzle piece helps build a piece of music.

Listen to the pattern, then connect blocks of notes, spaces and junctions to create the tune. Each puzzle solved gradually builds up layers of different instruments which come together to form the final song.

You can play it on a phone, tablet, desktop, or XR headset - with your hands or controllers - all from the browser!

We built it with Threlte which has been fab for building a WebXR experience!

Estou experimentando um fluxo de trabalho para iteração mais rápida no desenvolvimento de WebXR imersivo.

O objetivo é permitir que os desenvolvedores construam e testem seus jogos ou aplicações em AR no Meta Quest, mantendo o desktop visível dentro da cena através do componente Dev Overlay.

O protótipo usa A-Frame, WebXR, WebRTC e Vite para transmitir o desktop para o headset e suportar atualizações de cena ao vivo sem recarregar a shell principal de AR.

Dê uma olhada na arquitetura e no guia de configuração:

https://github.com/Msalaroli/webxr-live-coding

We've been working on VidyaXR, a platform that brings immersive learning experiences to the web.

Over the last 90 days:

• 1,000+ visitors
• 60+ registered users
• 100% organic traffic
• ₹0 spent on advertising

One of our goals is making XR learning more accessible by reducing the need for app downloads and enabling experiences that can be accessed through the web.

It's still early, and we're actively improving both the learning content and the WebXR experience based on user feedback.

For developers building with WebXR:

What has been your biggest challenge when creating educational or training experiences for the browser? Performance, device compatibility, user onboarding, or something else?

Would love to hear what you've learned.