When a dynamic scene shall be rendered fast, the easiest way is to lower the resolution during movement to make rendering faster, and higher up the resolution after scene becomes static again, as aligns with human vision perception.
ncastaneda02
Another trick to render dynamic scenes quickly is taking a low persistence approach, wherein only a segment of the scene is rendered at a time when it is "correct" and the rest of the screen left black. At a high refresh rate, this results in an imperceptible difference, but results in the hardware only having to do a fraction of the work at once by taking advantage of a fast refresh rate to allow for the compute to catch up.
akhilvemuri
A big issue with VR is the constant struggle between creating a powerful yet subtle experience. VR experiences require high-end hardware to render graphics and track movement in real-time. Upgrading the GPU, CPU, and memory of the computer can improve the overall performance and reduce latency. But also, having to wear big clunky hardware will reduce consumer interest and market value.
ZiqiShi-HMD
For decreasing the head tracking latency, we might be able to take advantage of the human body having internal latencies. Is it possible to win some time through BCI technologies? Recent research with EEG allows crude limb action reconstruction, but they didn't present data about latency and the accuracy is also pretty horrible (accuracy on the magnitude of 10cm). But I'm thinking if intrusive implants can achieve better results than EEG.
When a dynamic scene shall be rendered fast, the easiest way is to lower the resolution during movement to make rendering faster, and higher up the resolution after scene becomes static again, as aligns with human vision perception.
Another trick to render dynamic scenes quickly is taking a low persistence approach, wherein only a segment of the scene is rendered at a time when it is "correct" and the rest of the screen left black. At a high refresh rate, this results in an imperceptible difference, but results in the hardware only having to do a fraction of the work at once by taking advantage of a fast refresh rate to allow for the compute to catch up.
A big issue with VR is the constant struggle between creating a powerful yet subtle experience. VR experiences require high-end hardware to render graphics and track movement in real-time. Upgrading the GPU, CPU, and memory of the computer can improve the overall performance and reduce latency. But also, having to wear big clunky hardware will reduce consumer interest and market value.
For decreasing the head tracking latency, we might be able to take advantage of the human body having internal latencies. Is it possible to win some time through BCI technologies? Recent research with EEG allows crude limb action reconstruction, but they didn't present data about latency and the accuracy is also pretty horrible (accuracy on the magnitude of 10cm). But I'm thinking if intrusive implants can achieve better results than EEG.