2 ms·
You can hide the latency in very interesting ways. For example, oculus link encoded a fraction of the screen and starts sending that while the remainder is sti
by vlovich123 4y ago
You can hide the latency in very interesting ways. For example, oculus link encoded a fraction of the screen and starts sending that while the remainder is still encoding. That hides a non trivial part of the encoding latency. If you could reach further into the rendering pipeline (ie rendering the top 1/4 of the screen before the others) you could imagine reducing the effective latency even further by overlapping rendering with network transmission. It’s obviously tricky due to several reasons, but it’s not impossible (eg mobile games are rendered this way and mobile GPUs are particularly geared towards tile-based rendering). There are obviously limits to what you can do with reactivity to input but a lot of that can be mitigated with local prediction rendering.