Meta Quest Pro supports eye tracking and gaze rendering, but how much performance can it improve?
Eye Tracking Attention Rendering (ETFR) is a technology that renders the currently viewing display area at full resolution, and the rest is lower in resolution due to the release of performance, while additional performance can be used to improve the graphics fidelity of the application, or to improve the basic resolution.
Generally speaking, people don't notice the lower resolution of the edges, because the human eye can only see the high resolution of the center (fovea), which is why you can't read a page of text in full without moving your line of sight.
ETFR has long been considered VR "holy grail" because if GPU really only needs to render 3-degree field of view at full resolution, the performance advantage will reach 20 times, which will enable ultra-high resolution displays or incredibly detailed graphics. But in reality, to achieve this, perfect zero-delay eye tracking, an outrageously high display refresh rate and high-quality reconstruction algorithms are needed to prevent users from noticing flickering and glimmers.
Quest Pro is Meta's first headset with eye tracking function. The first generation of eye tracking technology has an end-to-end latency of about 50 milliseconds, and the display refresh rate is up to 90Hz. Therefore, its gaze point rendering is actually less than 20 times.
Since Oculus Go six years ago, the Meta headset has supported fixed foveal rendering (FFR)—rendering the edges of the lens at lower resolutions. In a presentation for developers this week, Meta detailed the exact performance benefits of eye tracking gaze point rendering (ETFR) and compared it to FFR.
, both types of gaze renderings are enabled by developers based on each application. Developers have three options to reduce the resolution of peripheral devices: Level 1, Level 2, and Level 3; in ETFR level 1, peripheral rendering pixels are reduced by 4 times, while in Level 3, rendering pixels are reduced by 16 times.
In addition, the exact performance advantages of gaze rendering also depend on the basic resolution of the application. The higher the resolution, the greater the performance cost savings.
In addition, Meta found that in performance testing applications, FFR saves 26% to 36% performance based on the fovea level at default resolution, while the new ETFR saves 33% to 45%.
However, at the default resolution of 1.5 times, the performance savings are higher, with FFR resolutions of 34% to 43%, and ETFR resolutions of 36% to 52%, which is up to 2 times more than not using the fovea, but this is only a small gain compared to FFR.
Of course, what is really important is that we don’t know yet: how significant is the ETFR level? So what is the performance compared to FFR? This requires comparison—rather than a given level of FFR with the same level of ETFR.
is worth mentioning that for Quest 2, FFR level 1 is not obvious at all, but level 3 has significant effects. Since the Quest Pro lens is sharper in the center and edges, FFR may be more eye-catching than ever before, giving ETFR an advantage.
In addition, for PSVR 2, the performance advantage of fovea rendering is greater, Sony claims that its FFR saves about 60% performance costs, while ETFR saves about 72%, which may be attributed to the very different GPU architectures of host and PC GPUs and the higher resolution compared to mobile GPUs; this may also be attributed to the differences in eye tracking technology, where Meta is based on internal features, while Sony uses Tobii technology.
Eye tracking features for both Quest Pro and PSVR 2 are optional for privacy reasons, however, disabling this feature also disables ETFR, so the application will have to fall back to FFR.
"93913 original content, please indicate the source when reprinting"
