kanetw

10 ms·

61.
by kanetw 4y ago
The 2 resistor on CC line thing is not spec compliant. E: you also need vRd sensing to determine the available current.
62.
by kanetw 4y ago
Absolutely. On top of that, I feel like USB-C is deliberately engineered to sell more ICs. Which wouldn't be that huge of an issue except as a result, all relevant ICs are sold out (or unsuitable). CC/PD use a custom transmission
63.
by kanetw 4y ago
Yep. If there's no port protection IC (and even then it might not catch it), if you unplug the USB-C connector you can potentially get 20V+ on the data lines. Goodbye data lines.
64.
by kanetw 4y ago
As a user, I liked USB-C. Until I had to implement this ridiculously overcomplicated garbage. To be spec-conform, you need at least 1 IC. You can theoretically hack a solution together with some resistors, but it's not spec conform. If
65.
by kanetw 4y ago
> Still probably should have a spec sheet, since your target demographic probably cares quite a bit more than average. I've suggested to George to add a more prominent spec sheet. It's there, but you need to scroll a bit or go
66.
by kanetw 4y ago
Not sure how other headsets do IPD, but with eye tracking it's pretty hard to get it wrong. If the eye's off center it's easily noticeable in the image.
67.
by kanetw 4y ago
https://uploadvr.com/samsung-wide-fov-patent/ VR purposes. This might be what was used in the Odyssey, come to think of it.
68.
by kanetw 4y ago
If Microsoft decides to give us access to the APIs to implement a window manager (however you'd do that on Windows...), possible. As is, would require a significant redesign of our approach. That being said, tethered mode will be Windo
69.
by kanetw 4y ago
You can gaze anywhere without significant quality loss (slight PPD reduction past 20 degrees but it's not really noticeable and it's uncomfortable to have your eyes that far away from center)
70.
by kanetw 4y ago
Adjusting IPD is really annoying at times. Probably one of the main reasons I pushed for auto IPD in our feature set.
71.
by kanetw 4y ago
Infinite focal distance isn't really doable with the space constraints you have in a headset. We tried, it massively increases complexity for no real gain.
72.
by kanetw 4y ago
We're 2448x2448 and use optics that magnify the center more than the edges, leading to even higher resolution (36 PPD, about equivalent to a 27" 1920x1080 monitor 60cm away)[1] [1] https://simulavr.com/blog/pp
73.
by kanetw 4y ago
About 1-2m in our design.
74.
by kanetw 4y ago
There's a patent by Samsung on it, but no actual product. There's always optics in front, so it's not immediately helpful. But a curved surface might simplify lens design.
75.
by kanetw 5y ago
I haven't gotten too deep into it myself, but the driver outputs a backlight PWM pulse a specific amount of time after a vsync. It's something I'll optimize at some point, but I have bigger problems to work on right now.
76.
by kanetw 5y ago
Not with this hardware.
77.
by kanetw 5y ago
The above formula only holds for diagonal FOV, which is higher than hFOV/vFOV. The formula for e.g. PPD in the h-direction, assuming uniform pixel distribution, would be hPPD = hRes / hFOV. With a uniform distribution, 2448px at 1
78.
by kanetw 5y ago
It's indistinguishable from reality at 60 PPD (pixels-per-degree) in the foveal region. We're at 35. Varjo has >60 PPD by having a high-res display for the foveal region and another for the periphery. It's doable with our
79.
by kanetw 5y ago
Yes, the compute pack has several DP outs and can be used without the headset.
80.
by kanetw 5y ago
I haven't turned my camera on in months, but I get your point.
81.
by kanetw 5y ago
Not for this version, but definitely for subsequent ones-- assuming we have the volumes to make AMD consider us.
82.
by kanetw 5y ago
Not yet using the prototypes (there's basically literally two samples of the displays we're using and we need them for dev) but once we have more of them produced, yeah, definitely. George used a Vive Pro for early software dev a
83.
by kanetw 5y ago
I'll post a through the lens image in a few weeks but this is equivalent to a 1080p monitor two feet away anywhere where you can direct your gaze. (Slight dropoff towards the edges but still over 30 PPD and 20um RMS spot size)
84.
by kanetw 5y ago
Correct. Visual-inertial SLAM using the AR passthrough cameras, with possible additional input from mmWave RADAR.
85.
by kanetw 5y ago
The focal point in our headset is on the order of ~2m. Infinite conjugate adds significant difficulty and basically nobody does this (we tried in our first iteration). But 2m is comfortably far, imo
86.
by kanetw 5y ago
AR mode is tricky, and it's one of the things I need full integration for. This will take a few months to get the sensors, FPGA, etc. sorted. So far with prototype units it was fine, but haven't tested for longer periods of time.
87.
by kanetw 5y ago
We have several external DP ports via USB-C. Web conferences... hm. Difficult to have some form of webcam with the headset on, but maybe there's a good way we'll discover down the line.
88.
by kanetw 5y ago
It's really incredibly hard to get a loan nowadays. And finding a VC that won't stick their nose in too much is also really hard. George knows more about it, but even an established company I'm familiar with could not get a l
89.
by kanetw 5y ago
Thank you! We want to avoid being bought at all costs :)
90.
by kanetw 5y ago
Swappable isn't planned for the v1, yeah. V2, maybe. Charge time should be about an hour (1C charging).

More ›