As I’ve been testing various Oculus Rift demos to prepare for our first virtual reality prototype, I noticed that one of the many differences with development for games for regular screens, is that you can’t show the same amount of information and text on the screen.
This limitation is mainly due to the low display resolution of the developer version of the Oculus Rift. These limitations remind me of my time developing for the Nintendo Wii – where working with standard-definition TVs meant having to support a worst-case resolution of 640×480.
The Oculus Rift DK1 (the first and at the time of writing only Development Kit), works with 640×800 per eye. And in fact, it’s worse than standard-definition TV: only 73% of the screen’s pixels are actually used, and the ones in the center are perceived bigger due to the lenses required for the Rift’s impressive viewing angle.
While discussing these limitations with our client, he simply replied:
Don’t worry about the text limitation by the current Oculus Rift. What we will build for, is the 4K version.
At first I thought having a 4K resolution (3840×2160, or 1920×2160 per eye) in a VR device would be only be possible in many years from now, because it would require such a high pixel density.
After doing some math, I was surprised that a 4K resolution Oculus Rift would be feasible with current technology:
- The screen size of the Oculus Rift DK1 is roughly 6.69×4.65 inch (measured on the outside), or 8.15 inch diagonally.
- The Rift’s 1280×800 resolution has a diagonal of 1509.44 pixels, resulting in a pixel density of 185 pixels per inch (ppi).
- A 4K resolution (with a diagonal of 4405.81 pixels) at this size would need to have a whopping pixel density of 540 ppi.
- The highest resolution iPads and iPhones have a pixel density of 326 ppi (60% of what would be needed for a 4K Rift).
- Looking at Wikipedia’s List of displays by pixel density, we learn that the HTC One, on the market since March 2013, has a 1920×1080 screen with a 468 ppi density. According to the same Wikipedia article, the Chinese electronics manufacturer Oppo Electronics would have a 2560×1440 screen in development of 538 ppi (rumoured to be on the market somewhere in 2014).
- A 1920×1080 screen at almost the right density doesn’t mean that a 4K resolution at this density would be possible. A few Google searches later, I found that Samsung would be working on phones with a 4K resolution and 560 ppi.
At first I was very sceptical about the low resolution of the Rift, but a bit of math later, I’m actually convinced this is technically feasible.
Whether the first consumer version of the Oculus Rift would be shipped with a 4K resolution screen is rather doubtful though, as Oculus’ Palmer Lucky stated that a 4K version would not be in development.
> only 73% of the screen’s pixels are actually used
It depends on the distance between the eyes and the lenses. You can find a multitude of sources with bigger images. Example : http://ctrlaltstudio.com/images/2013/08/Ahern-Welcome-Area-with-Oculus-Rift.png
> Oppo Electronics would have a 2560×1440 screen in development of 538 ppi (rumoured to be on the market somewhere in 2014)
The Vivo Xplay 3S has the same resolution, is already available for preorder and has already been reviewed : http://reviews.cnet.com/smartphones/vivo-xplay-3s/4505-6452_7-35834999.html
> A 4K resolution (with a diagonal of 4405.81 pixels) at this size would need to have a whopping pixel density of 540 ppi.
The pixel density of the Sony HMZ-T1 and the SMD ST-1080 is over 2000 ppi. It’s left to be seen if these displays can be produced at a higher size (<1" for now), but if there is a market I guess they'd find a way.
But I don't think there is a market for consumer 4K HMDs yet. There is one for pro HMDs though, and 4K is already available in this market (Sensics piSight at 2560×1600 per eye, ie. 5120×1600).
Thanks for these additions, Fredz.
The Sensics piSight seems to be very interesting. Their website also mentions they have a HMD (head mounted device) with 4200×2400 per eye, which is already a 4K resolution:
http://sensics.com/head-mounted-displays/technology/pisight-ultra-panoramic-hmds/overview-3/
It seems that these other devices (Sensics piSight, Sony HMZ-T1 and SMD ST-1080) don’t measure head rotation though. This could be added by simply adding an accelerometer and gyroscope (could be a phone, if it’s to play around with it).
Given the new specs for DK2 shown on this page:
http://www.roadtovr.com/oculus-rift-developer-kit-2-dk2-pre-order-release-date-specs-gdc-2014/
what changes to the current info in this article can be made?
I had a chance to test the DK2 at the GDC last week, and the main changes were the larger resolution and motion tracking. Regarding the resolution – this is already an improvement; however, the pixels are still noticeable (but just a bit smaller).
The DK2 is at 960×1080, while a 4K device would feature 1920×2160 per eye.
9th Jan 2014 Palmer said there is no 4k Rift in Development right now. Thats before facebook even approached them. After getting $2bn and infinite resources he has said :
CV2 just became CV1
and
now we can make our own custom hardware
According to AMD, you’d need a 8k by 4k resolution in order to achieve ‘photorealism’ in VR:
http://www.develop-online.net/news/vr-hardware-needs-to-be-200-times-more-powerful-to-achieve-photorealism/0201850
Not sure whether that’s per-eye or not, but it would at least require a 1080 ppi display. Yes, pixels per inch.
Good news though: Samsung announced to be releasing 880 ppi screens in 2015, bringing us very close. See:
http://www.extremetech.com/computing/170451-samsungs-2014-phones-will-feature-560-ppi-displays-880-ppi-in-2015-foldable-in-2016
Perhaps the question is: will VR fuel a new series of advancements in computing speed?
Urm Bernard, the amount of eyes do not really matter when you talk pixels per inch, it could be one eye or 4 eyes. It doesnt make a difference, just like a square meter will be a square meter even if nobody is measuring it.
@Ali: in this kind of VR headsets, we’re talking about a screen where the left side is used for the left eye, and the right side is used for the right eye, in order to achieve a stereoscopic 3D image. So you can consider the resolution per eye (which is the actual resolution you’re seeing), or the resolution of the display.
I re-read the relevant part in the source article and it seems they mean the total resolution, used for both eyes together, as they’re talking about the resolution that would be required for the screen.
Of course a bigger (hopefully not heavier) screen could also do the trick, bringing the technical feasibility of the photorealism the AMD guy was talking about a bit closer.
But to be honest, it’s feels old-school to me to make a comparison with photos when gauging the quality of a VR experience… I guess that depending on the resolution of the photo in ‘photorealism’, this has already been achieved :-).
More than 6 years after writing the orignal article, we’re still not at 4K for mainstream headsets. According to this article, the recently released Oculus Quest 2 has a resolution of 2K per eye, which would mean a pixel density of perhaps 270 ppi.
Meanwhile, Samsung and Stanford University researchers are moving the boundaries of display technology by borrowing advancements of solar panel technology within the field of nanosacle photonics. In a few years from now, dramatic improvements that bring display technology to a ppi of 10,000 ppi or more may be possible…