|
|
|
|
|
by basch
2649 days ago
|
|
Render the next 60 possible frames (and recursively the next 60 of each 60 to that frame), send them to the client, then based on controller input, it knows which frame to choose next. Use AI to determine the "most likely futures" and process those branch further. It could dynamically increase and decrease the "possible futures" sent depending on how fast a person is going and how many input choices they have at any given moment. Compression (combining like elements of different future frames) can happen strongly on frames further into the future that have more time to be decoded. |
|
"lol" is the best reply.