As I estimate there is impossible to use eye tracking for foveated rendering in CloudXR:
- best latency for client → server render → client is ~100ms
- eye movement (saccade) is upto 900°/s
→ up to 90° in one render cycle that is nearly the same as headset FOV
Or I am wrong and all latency problems can be solved in CloudXR (Latency Requirements for Foveated Rendering in Virtual Reality) ?