This just goes to show the rift vs vive argument is very much a subjective matter. However, objectively, the vives tracking technology is superior in design (and elegance) to the rifts. Though they may effectively produce the same result in 90% of cases, the vive has way more potential (tracking multiple people and objects, as its passive). For this alone IMHO the vive wins.
But of course, as there are many subjective factors to consider, so its going to be different for everyone. Having variety and choice is great for consumers though!
First off, I'm a huge fan of the lighthouse tracking system, and I'm looking forward to them opening it so I can use it in other projects. I do think it has some elegance to it.
That being said, it's not necessarily "superior in design". Because it uses a sweeping laser, it can't know where every tracking point is at the same time. It can't even measure each axis at the same time. The laser sweeps top to bottom and then side to side. For each tracking point, it records it's position in one axis at one point in time and the next axis at another point in time. If the tracking point has moved in between sweeps (which is likely), then you have two separate position measurements from two separate times that have to be combined with data from the IMUS to determine exact location. The Rift doesn't have this issue. The Rift records the location of all visible tracking points at the same time using an infrared camera.
It's important to keep in mind that Vive's lighthouse tracking system and Rift's constellation system only use the lasers and cameras to do drift correction. Most of the tracking is done by their IMUs (accelerators and gyroscopes), so the actual results you perceive will be very similar.
As you stated its not used for all tracking, but simply drift checking in pose estimation. So needing to update each axis simultaneously is not essential. The pose estimation is just updated with the latest values to correct for drift. At 60Hz this is sufficient, as the average movement speed of humans is likely not fast enough to produce noticeable problems.
The rift on the other hand has to do image processing. This requires significantly more processing power and is only as good as the resolution of the camera. The further you move from the camera, the less accurate it gets. The tracking device must also be tethered to the PC (via usb), limiting potential for adding multiple tracking cameras to improve accuracy. Also the more cameras you add, the more processing is required.
The vive does not have these issue - as long as the laser is in range, it will be as accurate as if you were right next to it. Adding more base units improves the accuracy at no extra processing cost. The design of the vive also allows pose estimation to be distributed. Ignoring occlusion issues in a small space, the lighthouse system could be used for tracking hundreds of devices.
What this essentially means, is that you can walk around a room with the vive and have a great experience [1]. While this might be possible with the rift, the experience is likely not going to be as good as the vive, and may be quite poor at times.
First off, I want to reiterate that I'm most excited about the lighthouse tracking system, it fits well with projects where I'd like to apply tracking, but my main objection is to your comment that it's an objectively superior design. All designs have pros and cons. Also, I realize I'm nit picking, but there's a couple inaccuracies in what you've said:
I'm not sure where you're getting the significantly more processing power; particularly if you're comparing to lighthouse. General image processing can add latency, but they've constrained the tracking problem, and it currently doesn't heavily utilize the CPU [1] . You can perform similar tracking cheaply with an ASIC like castAR does [2]. Lighthouse still has to perform sensor fusion with different time stamps for the tracker locations as I previously mentioned, which can also be processor intensive. I haven't done or seen any CPU usage comparisons between the two, but I believe they'd be comparable.
You are limited by the resolution of the camera, but you can still get subpixel accuracy when tracking. With lighthouse, you don't exactly get infinite resolution either, but I do think lighthouse takes the cake here. Once again though, it's the IMU that ultimately determines the accuracy.
Lighthouse being untethered is one of the more exciting things to me as it lends itself well to doing tracking in large event spaces.
The limiting factor for lighthouse isn't whether the laser is in range, but whether the LEDs are in range. In between each laser sweep, the IR LEDs flash, so that the tracking points know when to start timing [3]. I believe they've improved the distance of this flash in the consumer Vive, but they will still limit the distance.
Adding base stations does help improve accuracy, just as it does with constellation, but it won't improve precision. It only helps prevent issues with occlusion. However, adding more base stations won't improve the precision of the tracking device's location.
I believe the current lighthouse systems are still technically limited to two base stations, because of how they are time multiplexed. They're planning on frequency multiplexing which should allow this to grow without having to connect the base stations.
Also, lighthouse is susceptible to IR interference in ways that constellation isn't. They both utilize IR for their tracking, but since each tracker in the constellation system flashes a pattern, it's able to eliminate environmental IR. Lighthouse doesn't have a direct way to prevent IR interference that I'm aware of other than accounting for potential errors in their pose estimation.
Lighthouse also has moving parts, so there's a chance for mechanical failure. The motors are manufactured by Nidec [4] who are one of the largest manufacturers of hard drive motors, so they should be very high quality, but it is a potential point of failure.
Lastly, your statement "while this might be possible with the rift, the experience is likely not going to be as good as the Vive, and may be quite poor at times," seems to be entirely speculation. There are several tests with both single and multiple cameras that have performed room scale tracking exceptionally well. [5] [6] [7]
These are both great technologies that have their own strengths and weaknesses. I don't think there's one objective winner here. That being said, I'm looking forward to the updates to lighthouse that they've been talking about to make lighthouse viable for large event spaces, and I could even see a day where other IOT objects utilize lighthouse for tracking in ways that wouldn't be possible with constellation.
The rift has more steps in the tracking process. It has to track the dots of the currently displayed pattern. Transpose that into 3D space. Then it can perform triangulation.
The vive needs to only convert the time information to position relative to the base station. Then using the base station position, it can localise in 3D via the same/similar triangulation approach as the rift.
Hence, the rift requires significantly more processing power than the vive. This will likely introduce some latency, though its probably negligible to human perception. Furthermore, as the rift shows different patterns, it is possible, if moving fast enough for the patterns to move between frames, warping the dot patterns of the headset. This is a similar problem to the vive's sweeping. The major problem I see with the rift, is that adding more objects to be tracked only increases latency and complexity for tracking from a 2D image (more flashing patterns).
The lighthouse does not need to handle different timestamps for tracking. At each instant in time when new sensor data is received, the pose model is updated with the new sensor info (similar to a kalman filter). The pose model can be designed to update a single axis at a time though. There is only ~17ms between each axis update. You would have to be moving very fast to cause issues with tracking.
So, objectively the vives tracking is a superior design. Its elegant and simple. The vive and rift may perform similarly within the restrictions of their respective domains of use, such that 90% of the time people can't tell the difference. However, I would argue the vives domain is conservative. Start to move outside of those domains, the lighthouse system has much more potential and very few limitations that cant be reasonably/practically mitigated.
Both technologies have their advantages and disadvantages. Both allow tracking multiple objects (you need to add either LEDs or photodiodes to those objects) and adding more trackers (lighthouse towers or Constellation trackers). In practise I think for more cases they will be good enough, especially when the Oculus Touch controllers start shipping (they come with a 2nd tracking camera).
I've experienced not-so-solid tracking on the Oculus Rift CV1 with a single Constellation tracker when facing away from the tracker.
On the other hand, Tested.com reported in their HMD comparison that they have seen better tracking performance by Oculus than the HTC Vive. They are one of the few people seeing it that way, however - most reports indicate that the Vive tracking is superb.
Whichever headset you buy, if you are aware that it is a first generation product and can live with the high price, you will not be disappointed.
> Both technologies have their advantages and disadvantages.
I keep seeing this, but no one ever lists the Rift's advantages. Even in your post, you hedge on everything that might give the Rift leverage. At best it seems it will reach parity with the Vive when (if?) the Touch controllers ship, but even then I do think the Vive's fundamental tech is better designed.
I have a rift and a vive on order, the rift isn't comfortable at all with my glasses, where the Vive pre I've used is fine. It's easier to setup, but it also lacks controllers and the second sensor right now so it's not really a fair comparison.
I too wear glasses. The Vive is easier to put on over my glasses, but ultimately I find it less comfortable than the Rift once it is on. I find making comfort comparisons more than fair, I'm not sure I totally understand you last point.
But of course, as there are many subjective factors to consider, so its going to be different for everyone. Having variety and choice is great for consumers though!