22 Commits

Author SHA1 Message Date
Cameron Gutman
6c5ec3d2e9 Store the timestamp of the first packet received in the decode unit 2014-06-19 19:09:00 -07:00
Cameron Gutman
38423a9f37 Refactor the video decoding path so the DecoderRenderer handles pulling decode units instead of dedicating a separate thread for this 2014-06-19 18:28:48 -07:00
Cameron Gutman
0c8c108bd1 Add support for sending proper packet loss statistics for server-side bandwidth scaling 2014-05-11 18:49:20 -05:00
Cameron Gutman
92adbe0983 Update control protocol for GFE 2.0.1 2014-05-06 23:31:09 -04:00
Cameron Gutman
7e30d043eb Fix handling packet loss in the second of a split frame 2014-03-13 01:31:47 -04:00
Cameron Gutman
c8c7512600 Hack around an issue where data gets added after the frame is thrown away 2014-02-27 01:31:07 -05:00
Cameron Gutman
50e7deeb32 Merge branch 'master' into logs
Conflicts:
	src/com/limelight/nvstream/av/audio/AudioDepacketizer.java
	src/com/limelight/nvstream/av/video/VideoDepacketizer.java
	src/com/limelight/nvstream/control/ControlStream.java
2014-02-26 16:22:04 -05:00
Cameron Gutman
e60420cb2c Update depacketizer to do reference frame invalidation more like the official streamer. This should reduce the frequency of IDR requests by waiting for network stabilization before requesting the IDR frames. We still request IDR frames because reference frame invalidation still doesn't work well. 2014-02-26 12:12:06 -05:00
Cameron Gutman
4fbe93e62d Set flags on the decode units that indicate what type of data the frame contains 2014-02-26 01:00:17 -05:00
Cameron Gutman
bc2ca0b386 Increase resilience to packet loss. IDR frames are no longer requested if error correction data was lost. A maximum of one IDR frame is requested per corrupt frame. Error correction data is used to recover from the loss of a single-packet frame. 2014-02-24 12:54:03 -05:00
Cameron Gutman
63ee6ef79a Add support for direct submission of buffers to the renderers without a separate thread 2014-02-19 20:36:53 -05:00
Cameron Gutman
cdf634dc41 Display messages if we detect that the device is having issues streaming 2014-02-19 19:03:51 -05:00
Cameron Gutman
a96de39b28 Use packet flags to determine where frames end and begin instead of the packet index 2014-02-17 15:17:20 -05:00
Cameron Gutman
26809c4b6b Decode parameter set NALs with the slow path so the SPS fixup hack still works 2014-02-17 13:57:15 -05:00
Cameron Gutman
0cce5b021e New video depacketizer that runs in O(1) time 2014-02-17 13:39:18 -05:00
Iwan Timmer
ae18e00b13 Reassemble NAL's early 2014-02-03 00:30:44 +01:00
Aaron Neyer
f95cd60cfd Created LimeLog as a wrapper around javas logger 2014-01-12 20:23:18 -05:00
Cameron Gutman
6a695d2c72 Fix H264 decoding artifacts when a frame has fewer than 3 bytes of padding at the end 2014-01-11 14:27:46 -05:00
Cameron Gutman
042f67506c Remove the SPS hack from common, since it introduces compatibility problems for some devices 2014-01-10 00:27:35 -06:00
Cameron Gutman
b4a0f81eda Fixup bitstream_restriction_flag and max_dec_frame_buffering in the SPS to fix decoding latency issues on Tegra and the Raspberry Pi 2014-01-06 22:52:11 -06:00
Cameron Gutman
f1b4fdd8b0 Restore the old O(N) video depacketizer to fix artifacting. Add warnings for when the decode unit queue overflows. 2013-12-27 00:05:05 -05:00
Cameron Gutman
ce1494895e Commit of common limelight core with bindings based on HEAD of RenderScript-Renderer 2013-12-05 12:57:49 -05:00