HTC Vive and foveated rendering is now a thing

Associate
Joined
8 Jul 2007
Posts
215
https://uploadvr.com/7invensun-eye-tracker-for-vive/

If the requirement for rendering VR can take advantage of eye tracking, then the computer will not need to work so hard. It will render only the part that is being tracked by your eyes...

The advantage of this is (according to the linked article) that the minimum spec to run such a system is far lower than is currently the case with the technology used in the Vive. It's initially Vive only but by extension LGSteamVr and Facebook's Oculus and other VR headsets could potentially be developed for.

A welcome innovation, though hopefully with a lower pricetag, otherwise it seems less attractive to those looking to save a bit of money not upgrading their PC. It does however, suggest that the entry point for a PC supporting VR will get even lower than it is now, which has to be a good thing for adoption, uptake, development and innovation for VR generally.
 
Batman has some non-eye tracked foveated rendering techniques going on. A decent intermediate I would have thought, are the optics that good at the edges of the lenses anyway?
 
Batman has some non-eye tracked foveated rendering techniques going on. A decent intermediate I would have thought, are the optics that good at the edges of the lenses anyway?
I struggled to see any difference between the default fixed foveated rendering and full resolution rendering (pixel density at 220% for both). The only obvious difference was in the performance, so they're doing something right. :)
 
I struggled to see any difference between the default fixed foveated rendering and full resolution rendering (pixel density at 220% for both). The only obvious difference was in the performance, so they're doing something right. :)

Isn't that.. literally the point? Haha.

Performance increases the foveated rendering as the GPU doesn't have to work as hard. Rendering at full res across the whole lens is wasted effort.

Edit: think I read your post back in the wrong way.. thought you were expecting something else!
 
Last edited:
Isn't that.. literally the point? Haha.

Performance increases the foveated rendering as the GPU doesn't have to work as hard. Rendering at full res across the whole lens is wasted effort.

Edit: think I read your post back in the wrong way.. thought you were expecting something else!
With fixed foveated rendering (and the multi-res rendering) you would be able to see the quality take a dive if you looked at the edges of your field of view. Although I couldn't see it...probably because the focus isn't fantastic at the edges of my lenses in any case. :p

Edit - see here: http://support.wbgames.com/link/portal/24022/24028/Article/1487/Batman-Arkham-VR-PC-Graphics-Options

Proper foveated rendering with eye tracking - yes, that would be the point ;)
 
This and the wireless plug-in is going to make the Vive 2 an absolute beast of technology and a must have with them integrated into the headset. No reason they can't up the pixel density substantially and give us the res we're all dying for as well as increased FOV.
 
This and the wireless plug-in is going to make the Vive 2 an absolute beast of technology and a must have with them integrated into the headset. No reason they can't up the pixel density substantially and give us the res we're all dying for as well as increased FOV.

It's a guarantee that the next iteration will have a higher pixel density, the recently announced Acer and HP AR headsets both have 1440x1440 screens per eye with a 90hz refresh rate. By comparison the Rift and Vive are 1080x1200. At around 4k the screen door effect will be almost vanish, we just need to retain the high refresh rate of 90hz+.

But for that to happen we need HDMI 2.1 or DP 1.5? neither of which are available.
 
It's a guarantee that the next iteration will have a higher pixel density, the recently announced Acer and HP AR headsets both have 1440x1440 screens per eye with a 90hz refresh rate. By comparison the Rift and Vive are 1080x1200. At around 4k the screen door effect will be almost vanish, we just need to retain the high refresh rate of 90hz+.

But for that to happen we need HDMI 2.1 or DP 1.5? neither of which are available.

DP1.4 handles 120Hz 4k without the need for DSC. I wouldn't be against having 2 connectors, one for each eye. The question then would be whether the wireless connection could transmit at a similar rate which I have a hard time imagining.
 
4k over both displays would not equate to much of an improvement, it would need to be 4k per eye to really see a significant improvement to visual fidelity, that would need DP 1.5 as you'll be rendering at 8k, would likely need in excess of 100Gbit/s so wireless tech has a long way to go to allow that.
 
4k over both displays would not equate to much of an improvement, it would need to be 4k per eye to really see a significant improvement to visual fidelity, that would need DP 1.5 as you'll be rendering at 8k, would likely need in excess of 100Gbit/s so wireless tech has a long way to go to allow that.

Something for Lifi maybe, supposed to work to approx 220Gbit/s
 
4k over both displays would not equate to much of an improvement, it would need to be 4k per eye to really see a significant improvement to visual fidelity, that would need DP 1.5 as you'll be rendering at 8k, would likely need in excess of 100Gbit/s so wireless tech has a long way to go to allow that.

The suggestion from Abrash when he did a talk last year was that foveated rendering is also the key to successful implementation of wireless technology... the ability to only render the small spot you are looking at in high detail not only lowers the load on the computer, but also substantially reduces the bandwidth requirements for actually sending the images to the headset.
 
Back
Top Bottom