Early reviews have been mixed.
The big question mark is the tracking and how well it will work in the real world, particularly for the controllers. If they can pull it off, it will definitely be something to look for in the market. But if the tracking suffers, then the whole thing suffers.
That's pretty much the way I see it. The biggest problem right now is the lack of trustworthy information, and really, the number of similar-but-not-the-same headsets. I'm sure the basic implementation is consistent to maintain compatibility with MS's framework, but there are a few significant differences between models. The Samsung unit is definitely a step up from the others. Whether any of the differences are "under the hood" is anyone's guess at this point.
The headset tracking has been reported to be spot-on. The controller tracking, at least in this
, seems to be good. The inverse kinematic prediction seems to work pretty well in situations where only the internal sensors are able to be used, and at least one developer has indicated that the tools provide enough information to make things pretty seamless. When your hands are out of your view, precision positioning tends not to be very important anyway. I.e. in an archery game, it pretty much only needs to know that when you reach over and behind your head, you are after an arrow from your quiver. And when that arrow is loaded into the bow, and pulled back too close to your face to be tracked, it would lock it to your cheek position. While some might think this is "artificial", they probably haven't considered that many of the games they are playing already do this, if for no other reason than to allow players to focus on gameplay elements, rather than fiddly, mundane tasks.
The other question mark is going to be software availability. Reportedly it won't be compatible with SteamVR right out of the gate, which walls it off from a huge swath of content. If they can build in compatibility that would be a huge boon for these headsets. The last thing VR needs is yet another proprietary limitation when it comes to software.
Major reputable sources are indicating SteamVR compatibility "before the holidays", which is a bit nebulous. But it's already in beta so it's certainly going to happen. The interesting thing is that the MSMR will have compatibility with SteamVR and the MS library, but the MS side of things will be walled off for users of other headsets. But yes, I think the lack of available content is seriously stalling adoption at the moment, unless you just want a VR version of windows10, and to play virtual doll-house (although the built-in big-screen media room looks like a nice feature.)
The thing that really prompted me to check this out is the resolution bump. Previous offerings are 1080x1200 per eye, while the new Samsung's AMOLED is 1440x1600 per eye. The difference seems marginal when looking at these numbers, but it comes out to an extra ~1,000,000 pixels per eye, which is almost twice the resolution. This means SDE will likely be virtually eliminated, distance rendered objects sharper and text much more readable.
MS seems to be doing some trickery to get the system requirements down to handle higher resolutions as well. One VR "guru", while testing the lowest-end model, indicated that he noticed a rendering artifact which might be telling. He noted that during rapid head movements, a black bar would appear on the periphery of his vision, at the side opposite of the direction he was turning, and the full image would return when he stopped. While this may be something to help out the "queasy crowd", it could very well be a simplified foveated rendering technique. It's possible that the software is monitoring GPU and system loads, and applying this when necessary to reduce the area needing to be rendered. Or, it could have just been a pre-release glitch
But it does seem like a reasonable approach.