Samsung has taken the lead in the race for mixed reality (MR devices), unveiling an upcoming headset based on the newly introduced Android XR. Known internally as “Project Moohan,” Samsung anticipates the device will be available to consumers by 2025. I recently had the chance to test out an early prototype, providing some first impressions.
At this stage, neither Samsung nor Google has provided detailed specifications for the headset, such as its resolution, weight, field-of-view, or price. Unfortunately, I was also restricted from capturing any photos or videos during my demonstration. All we can rely on for now is an official image from the companies.
Imagine merging the capabilities of the Quest with those of Vision Pro: that gives you a pretty solid idea of what Project Moohan brings to the table. And this isn’t just a playful comparison. Even a quick glance at the headset reveals its significant design influences from Vision Pro—from the color scheme to the placement of buttons and even the calibration process. It’s clear that the team was aware of the competitive products already available.
On the software side, if I tasked you with combining Horizon OS and VisionOS, and you came up with Android XR, I’d say you hit the nail on the head. It’s striking how much this collaboration feels like a blend of the world’s leading headset platforms.
However, don’t get me wrong; this isn’t about claiming someone copied another. In tech, borrowing efficient ideas and design aspects is common, with each iteration potentially refining the original concept. If Android XR, along with Project Moohan, manages to capture the good and steer clear of the bad, developers and users stand to gain a lot.
And from what I’ve seen, many of the positive elements are indeed present.
Exploring Samsung’s Project Moohan Android XR Headset
Looking at the hardware, Project Moohan is undeniably sleek. It reflects the ‘goggle’ style associated with Vision Pro, but instead of the less comfortable soft strap (which often needs third-party upgrades), Samsung opted for a sturdy strap with an adjustable dial, reminiscent of Quest Pro’s ergonomic design. This allows for an open-peripheral setup, excellent for augmented reality experiences. For those wanting a more immersive environment, detachable magnetic blinders, akin to those on Quest Pro, are included.
Interestingly, while Project Moohan’s design—down to button placements—is reminiscent of Vision Pro, it lacks an external display feature to show the user’s eyes, something Vision Pro is known for (despite the debate on its usefulness). Personally, I find this feature desirable and feel its absence on Project Moohan.
Although still labeled a prototype, Samsung has confirmed that the headset runs on a Snapdragon XR2+ Gen 2 processor, which is a step up from the chips in the Quest 3 and Quest 3S.
During my demo, a few specs became apparent. The headset uses pancake lenses with automatic IPD adjustments, thanks to built-in eye-tracking. Initially, the field-of-view seemed narrower than that of Quest 3 or Vision Pro, but I need to experiment with different forehead pad options (which are included) to ascertain if eye positioning affects the field-of-view’s extent.
The current field-of-view feels slightly limited, striking a decent balance between immersive experience and brightness affecting the peripheral clarity. This might improve by adjusting the lenses closer to the eyes, but for now, Meta’s Quest 3 appears to take the lead, followed by Vision Pro, with Project Moohan slightly trailing.
Samsung will introduce unique controllers for Project Moohan, though I haven’t had a chance to handle them just yet. There’s also a decision pending on whether these will be sold as part of the package or separately.
For the testing, it was all about hand and eye-tracking inputs; the software feels like a compelling mix of Horizon OS and VisionOS. You can use raycast cursors like in Horizon OS, or opt for eye and pinch gestures seen in VisionOS. There’s even a neat integration of downward-facing cameras allowing comfortable use of pinch gestures while your hands are resting in your lap.
Upon wearing the headset, my most notable observation was the clarity of my hands in the virtual space. The passthrough cameras projected a sharper image than what Quest 3 offers, with less blur than observed in Vision Pro, albeit in optimal lighting conditions. However, while my hands appeared crisp, things further out seemed less defined, suggesting a focus at arm’s length.
A Deeper Dive into Android XR
Android XR stands as an intuitive blend of Horizon OS and VisionOS. Its user interface, featuring app icons over a transparent background akin to Vision Pro, aligns with a familiar home screen setup. With simple look-and-pinch gestures, apps open in floating panels, and navigating through them feels natural.
The system windows mimic those from Horizon OS more closely than VisionOS, sporting mostly opaque backgrounds and allowing adjustments by interacting with invisible frames surrounding each panel.
Android XR supports fully immersive content too. For instance, I experienced a VR adaptation of Google Maps, eerily similar to Google Earth VR, where you can explore any location globally, view 3D depictions of major cities, dive into Street View, and now even experience volumetric captures of interiors.
The Google’s Gaussian splat technique for interior captures produces decent yet not photogrammetry-level results. Yet, they managed to run seamlessly on the device, and I was informed that sharpness will continue to improve over time.
On Android XR, Google Photos allows automatic 3D conversion of any 2D photo or video from your library. Initial review suggests these conversions are strikingly effective, matching the quality of counterparts on Vision Pro.
YouTube on Android XR is also getting an upgrade. Beyond standard flatscreen viewing on a large, curved setup, you can also explore 180, 360, and 3D videos in the vast existing archive. While not all are high quality, it’s a nice touch for those valuing diverse media experiences on the platform.
Google demonstrated a YouTube video initially shot in 2D but converted to 3D for headset viewing. The conversion was noteworthy, bearing resemblance to the same feature in Google Photos’ tech. Whether this became a choice for content creators or an automatic process by YouTube remains to be seen.
Gemini and Its Power in Android XR
Where Android XR and Project Moohan truly shine is in their use of conversational AI, something that currently places them ahead of the competition. Google’s AI agent, Gemini (specifically the ‘Project Astra’ variant), boasts capabilities distinctly superior to those found in existing headsets.
Gemini can be activated from the home screen and maintains an awareness of both the real and virtual environments you’re interacting with. This constant perception lends it a smarter and more integrated feel than the AI agents in other headsets like Vision Pro or Quest.
While Vision Pro relies on a more task-simple Siri, and Meta’s Quest offers an AI agent with limitations, Gemini excels with real-time, low-framerate visual awareness, avoiding unnatural pauses during inquiries.
Gemini’s 10-minute rolling memory and comprehension of past conversations and events amplify its contextual understanding. During a demo, I challenged it with contextual questions, only to be impressed by its adeptness at providing correct answers.
When I requested translations of a Spanish sign to English, followed by asking for French phrases from another already in French, Gemini tackled these tasks, offering precise translations and even using appropriate accents where needed.
Remembering our dialogue, Gemini flawlessly recalled the signs I referred to. Such comprehension has historically been a complex challenge for AI, but Gemini navigates it impressively. Moreover, the AI not only answers general inquiries but also controls headset features, enhancing the immersive experience.
For instance, saying, “take me to the Eiffel Tower,” launches an immersive Google Maps view, all while maintaining a conversational flow, answering specifics like the structure’s height or construction date.
Gemini can also facilitate specific YouTube video suggestions, related to questions asked in a virtual setting. Such integrated assistant functionalities further illustrate its potential.
Moving forward, Gemini’s introduction in Android XR is the best AI integration I’ve experienced on a headset. Yet, Apple and Meta are undoubtedly on similar paths, so Samsung’s lead in AI could face challenges as the technology evolves.
While Gemini enhances the productivity aspect of Project Moohan, its ultimate potential may lie in more compact wearables, such as the smartglasses I also explored. But that’s a story for another time.
