Sony VR Headset/Project Morpheus/PlayStation VR
Finally! An advantage for elf
I guess android is a logical choice, it's by far the most mature OS for a modern ARM SoC and has very active development of all necessary drivers, it's easy development too. They can also take whatever they need from the entire project, it could be only the linux kernel and drivers, with none of the graphical environment or services.
I'm more curious about Sony open sourcing some of their own stuff in the morpheus bridge package. They wouldn't open source anything unless they either want to (PC port?), or are required to (licence requirements?).
http://doc.dl.playstation.net/doc/c.../psvr-oss/morpheus_bridge-PSVR-02.000.tar.bz2
It looks like the proprietary USB devices interfaces. Why open source this? I think it's either for a PC port, or maybe it's a license requirement because the kernel is monolithic (but it's GPLv2 here, while Android kernel is Apache Licence, I don't know how mixed licences work)
i cant find it. but not all android projects must have their source released right?
even on android phones, its only Sony Xperia Z lines that consistently upload their source
Where did you get those?
From your link... If you click on a GPLv2 package, the preface paragraph is saying "source code available here".
They list the linux kernel as a separate package because it's not the same licensing as Android.
http://www.roadtovr.com/playstation-vr-psvr-teardown-disassembly-samsung-display-lenses/
RoadToVR hypothesis doesn't make much sense, it's not that complicated. The white parts would be the metal around the oled elements.
*cough* I called this a year ago *cough*
OLED grid layouts are designed to maximize sub-pixel density, and the ideal situation is regular tiling, where the pixels are the same distance from all neighbors. It leaves only 2 choices on a flat surface: square and hex. (diamond pentile is simply a square grid rotated). For RGB it has to be hex, because it needs to divide the pixels in groups of 3. Orthogonal square or diamond is great for RGBG, but can't work with RGB.
RoadToVR hypothesis doesn't make much sense, it's not that complicated. The white parts would be the metal around the oled elements.
*cough* I called this a year ago *cough*
Still wondering what pixel structure they used, and who makes the screen, because I can't find any oled striped RGB pattern at a high enough density. True stripes are only used for TVs, samsung dropped the s-stripe pattern years ago, and never reached 300ppi. The diamond matrix really unlocked the density, but it's probably not adaptable to RGB because it's orthogonal.
I made dis...
OLED grid layouts are designed to maximize sub-pixel density, and the ideal situation is regular tiling, where the pixels are the same distance from all neighbors. It leaves only 2 choices on a flat surface: square and hex. (diamond pentile is simply a square grid rotated). For RGB it has to be hex, because it needs to divide the pixels in groups of 3. Orthogonal square or diamond is great for RGBG, but can't work with RGB.
Last edited:
I tried to use my Lumia 950 XL for VR as it comes with a 1440p screen, but it has more screendoor than the PSVR ... Pixels are quite clearly visible on it. Somehow it almost seems to stand out a little even when looking at a white background in 2D, despite being over 500ppi.
2D photos look great though, and the combo with a really good camera and strong cpu/gpu makes it a very impressive phone. Just a handful of apps missing from making me throw away my iPhone 6 plus really, which is impressive.
2D photos look great though, and the combo with a really good camera and strong cpu/gpu makes it a very impressive phone. Just a handful of apps missing from making me throw away my iPhone 6 plus really, which is impressive.
I'm wondering if lens design is also contributing to the lack of visible screendoor in a way that is not possibly with pentile. If they could limit the resolving power (maybe an "imperfect" profile by design?), it would remove the black space without losing resolution in any channel, since all 3 colors would react the same way. If you try an optical filter (defocus) on the above pattern, you can get a relatively uniform white, but that doesn't work with pentile since the colors are not the same pattern and can't be blended with simply reducing the optical resolving power.
It all depends whether you have to use mass manufactured display panels originally intended for information/data displays with sunlight readibility like the ones they use now.
With a diffusing layer above the displaced subpixels, you trade off "modulation transfer function" , it will result in pixel overlap and even lower effective resolution.
A better alternative is time-sequential color , because the brightness requirement is very low, the different colors can be very-very closely spaced in time, with 90% MTF and no subpixels in space but time (no gaps are evident).
See :
http://www.kguttag.com/2016/10/13/n...or-resolution-fov-brightness-and-eyeboxpupil/
etc.
Remember that the displays for VR have a need for very brief image persistence to avoid motion artifacts on the retina, and preferably a global update to ensure that for the ~2ms that the frame is actually visible you have a complete static image that's not exhibiting any scan-out induced oddities.
Last edited:
psvr screen do have screendoor but its diagonal. when you see it, you can easily keep seeing it again when you look for it. But if you not looking for it, only the mura that are easily seen.
sony also said they specifically designed the pixel to be closer together.
my 1080p LG G Pro 2 actually have much better image quality (zero mura, vibrant colors) than PSVR but the screen door is easier to see. Despite on cardboard it got compressed to much lower FOV
sony also said they specifically designed the pixel to be closer together.
my 1080p LG G Pro 2 actually have much better image quality (zero mura, vibrant colors) than PSVR but the screen door is easier to see. Despite on cardboard it got compressed to much lower FOV
D
Deleted member 13524
Guest
http://wccftech.com/crytek-details-robinson-journey-ps4-pro-enhancements-impressed-sonys-ps-vr/
Perhaps a confirmation that some games - robinson included - are not being rendered at the native 960*1080 per eye.
Unless they're downscaling / using SSAA..
Perhaps a confirmation that some games - robinson included - are not being rendered at the native 960*1080 per eye.
Unless they're downscaling / using SSAA..
Sigfried1977
Legend
I don't think there are any games rendering below 960+1080 per eye at this point, including Driveclub. It's just that in Driveclub, you're naturally focusing on far away points on the road. In most other games you're meant to look at nearby details. The cockpits look marvelous and relatively clean, whereas the road in front of you gets hit hard by the low pixel density of the HMD. There's various methods of anti-aliasing, though. In the case of Windlands and Playroom VR, possibly super-sampling. Windlands in particular reaps tremendous dividends from its pristine image quality as it's another game where details in the distance actually matter.
Robinson looks very clean running on stock first gen PS4 hardware as well, by the way.
Robinson looks very clean running on stock first gen PS4 hardware as well, by the way.
Anyone tried Eagle Flight yet, is it any good? Is it very jaggy and blurry in further distances?
Sigfried1977
Legend
It looks really good. Not quite sure how I feel about the controls, though. You're steering the bird with your head movements only (I'd have prefered regular controls + free head movement). You also cannot disable any of the comfort features like the black vignette that closes in on your vision whenever you make a turn or fly very close to obstacles. It's really comfortable, though. I don't think anyone in our office felt bad after trying it.
Wow, just realized that Robinson The Journey is literally half the price in the US ($40 vs €70). What The Fuck.
Wow, just realized that Robinson The Journey is literally half the price in the US ($40 vs €70). What The Fuck.
Similar threads
