摘要:
Systems and methods are provided for reducing and controlling playback latency in an unmanaged, buffered data network. A delay cost function is determined, the function representing the effect of playback latency on end user experience. An encoder transmits audiovisual data through the network to a client device. Network latency is measured, and the delay cost function is evaluated to establish an encoding bitrate for the encoder. The encoding of the audiovisual data is altered in response to dynamic network conditions, thereby controlling end-to-end playback latency of the system, which is represented by the playout length of data buffered between the encoder and the client device.
摘要:
Systems and methods are provided for reducing and controlling playback latency in an unmanaged, buffered data network. A delay cost function is determined, the function representing the effect of playback latency on end user experience. An encoder transmits audiovisual data through the network to a client device. Network latency is measured, and the delay cost function is evaluated to establish an encoding bitrate for the encoder. The encoding of the audiovisual data is altered in response to dynamic network conditions, thereby controlling end-to-end playback latency of the system, which is represented by the playout length of data buffered between the encoder and the client device.
摘要:
System and methods are provided to cache encoded graphical objects that may be subsequently combined with other encoded video data to form a data stream decodable by a client device according to a format specification. Paint instructions relating to a graphical object are sent from a layout engine to a rendering library. A shim intercepts these instructions and determines whether the graphical object already has been rendered and encoded. If so, a cached copy of the object is transmitted to the client device. If not, the shim transparently passes the instructions to the rendering library, and the object is rendered, encoded, and cached. Hash values are used for efficiency. Methods are disclosed to detect and cache animations, and to cut and splice cached objects into encoded video data.
摘要:
System and methods are provided to cache encoded graphical objects that may be subsequently combined with other encoded video data to form a data stream decodable by a client device according to a format specification. Paint instructions relating to a graphical object are sent from a layout engine to a rendering library. A shim intercepts these instructions and determines whether the graphical object already has been rendered and encoded. If so, a cached copy of the object is transmitted to the client device. If not, the shim transparently passes the instructions to the rendering library, and the object is rendered, encoded, and cached. Hash values are used for efficiency. Methods are disclosed to detect and cache animations, and to cut and splice cached objects into encoded video data.