I was wondering how much latency was acceptable for eye tracking for foveated rendering and found this paper (https://research.nvidia.com/publication/2017-09_Latency-Requirements-for) that says <40ms is ideal.
arjunsrinivasan1997
What are some techniques used to actually track the gaze of the user?
Carpetfizz
I went to a talk held by an engineer at an AR company. They use cameras that look at each eyeball and track their movement at a very high frame rate. Vision science has models that map the position of an eyeball to their gaze, which is then fed into higher level systems like this one.
youtuyy
I think foveate rendering is a hot research topic in recent years because of the growing popularity of these head-mounted displays. It requires good gaze detection. Regard of the eye model thing in the previous comments, there are multiple modeled eye based on population average of clinical measurements . However, a better result could come from data from individuals, but it would be hard to realize.
jackcsullivan
This concept is also apparent in non VR 3d video games, where the background and areas far away from the user use lower resolutions.
/slides/slide-25.jpg)
I was wondering how much latency was acceptable for eye tracking for foveated rendering and found this paper (https://research.nvidia.com/publication/2017-09_Latency-Requirements-for) that says <40ms is ideal.
What are some techniques used to actually track the gaze of the user?
I went to a talk held by an engineer at an AR company. They use cameras that look at each eyeball and track their movement at a very high frame rate. Vision science has models that map the position of an eyeball to their gaze, which is then fed into higher level systems like this one.
I think foveate rendering is a hot research topic in recent years because of the growing popularity of these head-mounted displays. It requires good gaze detection. Regard of the eye model thing in the previous comments, there are multiple modeled eye based on population average of clinical measurements . However, a better result could come from data from individuals, but it would be hard to realize.
This concept is also apparent in non VR 3d video games, where the background and areas far away from the user use lower resolutions.