Samsung’s Android XR Headset: A Unique Blend of Quest and Vision Pro with a Distinct Edge

Samsung has made a splash by being the first to unveil a new mixed reality headset based on the freshly introduced Android XR, dubbed “Project Moohan.” This eagerly anticipated device is set to hit the consumer market by 2025, and I had the opportunity to experience an early version firsthand.

I should mention that Samsung and Google are keeping their cards close to the chest regarding specific details such as the headset’s resolution, weight, field-of-view, or price. During my demonstration, I was unable to take any photos or videos, leaving us with only an official image for now.

If you’re imagining Project Moohan as a blend of the Quest and Vision Pro, you wouldn’t be far off. It’s not just an educated guess—its design overtly echoes that of Vision Pro. From the color palette to the button arrangement and even the calibration process, all these elements reveal its awareness of what’s already out there.

On the software front, Project Moohan’s operating system could easily be described as a marriage between Horizon OS and VisionOS. In crafting Android XR, it seems they perfectly executed this blend.

Yet, this isn’t about accusing anyone of copying. In tech, borrowing and enhancing ideas are par for the course. As long as Android XR and Project Moohan cherry-pick the strengths of their peers while sidestepping the weaknesses, both developers and users stand to gain.

Thankfully, a lot of those positive aspects have made their way into Project Moohan.

Exploring the Samsung Project Moohan Android XR Headset

Starting with the hardware, you can’t help but appreciate the chic design of Project Moohan. It mirrors the ‘goggles’ aesthetic of Vision Pro, albeit with improvements. For instance, where Vision Pro’s straps might need third-party adjustments to avoid irritation, Samsung opts for a solid strap with a convenient tightening dial, evoking the ergonomics of Quest Pro. This headset’s open-peripheral design is ideal for AR applications. And, akin to Quest Pro, it comes with magnetic snap-on blinders for those seeking a more immersive experience.

The similarities to Vision Pro don’t end with looks—some button placements and their design elements also echo it. However, Project Moohan doesn’t feature an outward-facing display to show the wearer’s eyes. Although some critique Vision Pro’s ‘EyeSight’ feature, I personally think it’s quite beneficial, and I hope to see it eventually integrated into Project Moohan. Seeing someone’s eyes creates a sense of connection which is missing without it.

Samsung is reticent about the headset’s technological specs, labeling it as a prototype. We know, however, that it’s powered by a Snapdragon XR2+ Gen 2 processor, a beefier upgrade from what’s seen in the Quest 3 and Quest 3S.

From my hands-on experience, I discerned some specifics. For one, it boasts pancake lenses with automatic IPD adjustment through integrated eye-tracking. Although the field-of-view seemed narrower compared to Quest 3 or Vision Pro, I’m reserving judgment until I can experiment with different forehead pad options, which might place my eyes nearer to the lenses for a broader view.

As it stands, the field-of-view may be somewhat tight, yet it remains sufficiently immersive. Meanwhile, the sweet spot feels constrained by a brightness drop-off towards the display’s periphery. Again, this could improve with the lenses closer to my eyes, but as things are now, Meta’s Quest 3 leads, followed closely by Vision Pro, with Project Moohan trailing slightly.

While Samsung has confirmed Project Moohan will include controllers of its own, I didn’t interact with them during my test. It’s still undecided if these controllers will come bundled with the headset or sold as extras.

During my test, I relied entirely on hand-tracking and eye-tracking inputs, reminiscent of systems in Horizon OS and VisionOS. You can opt for raycast cursors like Horizon OS, or go for eye+pinch controls as seen in VisionOS. Samsung’s headset also features downward-facing cameras, enabling pinch detection even when your hands rest in your lap.

Upon donning the headset, I was struck by how sharply my hands appeared in the display. From what I recall, the passthrough visuals from the cameras seemed crisper than Quest 3, with less motion blur than Vision Pro—keeping in mind this was in ideal lighting. It seemed as though the cameras might be calibrated for focusing at an arm’s length.

Continuing on to Android XR

Turning to Android XR’s software, it’s an instant mix of Horizon OS and VisionOS. You’ll spot a home screen akin to Vision Pro’s, populated with app icons overlaying a see-through background. Just look and pinch to open an app, which appears as a floating panel—or multiple panels. Opening the home screen involves the same simple gesture: look at your palm and pinch.

The system windows themselves are closer in appearance to those of Horizon OS than VisionOS, sporting primarily opaque backgrounds with the convenience of repositioning them by touching an invisible frame encircling the panel.

Android XR also accommodates fully immersive VR content. I experienced a VR rendition of Google Maps, echoing Google Earth VR, where you can globe-trot, exploring places with 3D cityscapes, Street View imagery, and fresh interior space volumetric captures.

While Street View remains monoscopic, the volumetric captures are rendered in real-time for exploration. Google describes this as a gaussian splat solution, though it’s unclear if this relies on existing Google Maps photography or demands a new scan. It wasn’t as detailed as photogrammetry, but was quite neat regardless. Notably, the data is processed directly on the device, not streamed, with further clarity improvements anticipated.

Google Photos has received an Android XR upgrade, allowing automatic conversion of 2D photos or videos into 3D format. During my short demo, these conversions were stunning, akin to Vision Pro’s feature in terms of quality.

YouTube also taps into Android XR’s capabilities, offering not just traditional flat content on expansive, rounded screens but also its existing catalog of 180, 360, and 3D videos. Although not all options boast high fidelity, they’re actively supported and sure to grow as more headsets come online.

A YouTube clip originally shot in 2D received an automatic 3D makeover for headset viewing, delivering a quality akin to Google Photos’ 3D transformation tech. It remains unclear if this process requires creator consent or if it’s an automatic feature. No doubt more specifics will surface.

A Key Highlight (for now)

When it comes to Android XR and Project Moohan, it seems like Google has conjured a version of what’s currently available with thoughtful enhancements. Their standout feature is decisively Gemini, Google’s sophisticated AI agent.

Incorporating Gemini, specifically the variant known as ‘Project Astra,’ users can interact with it directly from the home screen. Not only does it listen, but it simultaneously sees both the real and virtual worlds around you—constantly. This continuous awareness leads to a more nuanced, integrated, and conversational AI experience compared to other headsets out there.

Apple’s Vision Pro has Siri, but it primarily revolves around one-off tasks and lacks conversational depth. Meanwhile, Quest’s experimental Meta AI agent can hear and see the real world, but not the virtual one, leading to a noticeable gap. Although improvements are planned for Meta, at present, you need to explicitly query your environment and await response after it takes a ‘snapshot.’

Gemini differentiates itself by effectively offering a low-framerate video feed of your surroundings, ensuring no awkward pauses and removing the need for direct visual focus related to questions.

Moreover, Gemini on Android XR possesses a rolling 10-minute memory that retains key conversation details. This means you can reference prior discussions or visual cues. In a standard AI demonstration, such as examining objects in a room, it handsomely navigated through trick questions.

For instance, I asked for a translation of a Spanish sign into English, which it promptly provided. Then, testing it further, I requested a translation of another sign into French, knowing it was already in French. Gemini smoothly acknowledged this, adding the translation in a French accent.

Continuing the conversation, I returned to previous signs I had asked about, making inquiries like “what did that last sign say?” Gemini accurately recalled what we’d discussed, demonstrating an impressive retention level.

On Android XR, Gemini effortlessly surpasses its predecessors in headsets—or even Meta’s current smartglasses on the market. Though both Apple and Meta are likely pursuing similar advancements, Google’s clear advantage may not hold forever.

The true potential of Gemini may ultimately shine in everyday smartglasses, which I had the chance to explore. But that’s a story for another time.