Actually, on our own AR devices, JSAR uses OpenXR to enable mixed rendering of web content and Unity scenes together. JSAR itself is a browser engine library, and it currently works in any AOSP + GLES3 environment (which should cover most devices except Vision Pro and Windows). However, I haven’t built a full browser app specifically for mainstream VR devices like the Meta Quest, as the project’s main focus right now is on implementing missing web standards.
Currently, JSAR supports rendering WebXR content along with basic Input Sources (including Hand Tracking on XR devices).
Thank you for the recommendation! After having experienced RP1, it appears to be a large-scale web spatial application. I believe JSAR could potentially integrate other WebXR content within environments like RP1's expansive space - which would be quite fascinating. However, to run within JSAR, RP1 would need to support OVR_multiview rendering.
You're absolutely right - the current situation of this project can indeed be confusing. Thank you for bringing this up! Would you mind sharing which XR devices you'd ideally like to see JSAR support? I'd be happy to prioritize building a compatible version for your preferred platform in my roadmap.
Hey, If you have a Rokid ARLite, it comes with JSAR by default and can be used directly. For other AOSP-based devices, such as the Meta Quest, you may need to use the JSAR SDK to develop a new Unity applications yourself: https://m-creativelab.github.io/jsar-runtime/manual/referenc...
Yea, it’s primarily designed for headset or AR glasses, but it can also function as a special web browser for desktop platforms (e.g., macOS). By default, the window is a 3D space where you interact with web content (both 2D & 3D).
JSAR itself is a browser engine, so it doesn't depend on other browsers' implementations (JSAR follows backward-compatible design principles, meaning it won't modify any existing Web standard implementations but rather extends capabilities based on current standards). I've personally integrated it into an AOSP + Unity environment (with underlying GLES rendering), and anyone can also port JSAR to other GLES-based scenarios.
In short, JSAR is designed for XR devices like Apple Vision Pro and Quest (though it currently only supports AOSP devices, with a simulator available for macOS). Its most important feature is rendering multiple WebXR applications in the same scene while coexisting with the host environment (e.g., Unity or Unreal Engine). Each WebXR app can use different tech stacks like Three.js or Babylon.js.
Additionally, JSAR implements spatialized HTML and CSS, rendering each DOM element as a 3D object rather than flattening everything into a single 2D texture. This opens up new possibilities for web developers.
Yes, I do want JSAR to progressively align with web standards and improve WPT compatibility (to match engines like WebKit), but that’s not the primary goal. For JSAR, the focus is enabling web developers to incrementally build spatial web applications—smoothly, like how Apple developers work with SwiftUI.
As for the website, it hasn’t been updated in a while and was never optimized for mobile (since this is largely a solo project, and my priority is coding). I’ve moved the documentation to the repo: https://github.com/M-CreativeLab/jsar-runtime/blob/main/docs..., you can browse it directly on GitHub mobile.