智能化下的产物,无人棋牌室
GStreamer 团队很高兴的发布了 1.6.0 版本,该版本主要值得关注的改进包括:
- Stereoscopic 3D and multiview video support
- Trick mode API for key-frame only fast-forward/fast-reverse playback etc.
- Improved DTS (decoding timestamp) vs. PTS (presentation timestamp) handling to account for negative DTS
- New GstVideoConverter API for more optimised and more correct conversion of raw video frames between all supported formats, with rescaling
- v4l2src now supports renegotiation
- v4l2transform can now do scaling
- V4L2 Element now report Colorimetry properly
- Easier chunked recording of MP4, Matroska, Ogg, MPEG-TS: new splitmuxsink and multifilesink improvements
- Content Protection signalling API and Common Encryption (CENC) support for DASH/MP4
- Many adaptive streaming (DASH, HLS and MSS) improvements
- New PTP and NTP network client clocks and better remote clock tracking stability
- High-quality text subtitle overlay at display resolutions with glimagesink or gtkglsink
- RECORD support for the GStreamer RTSP Server
- Retransmissions (RTX) support in RTSP server and client
- RTSP seeking support in client and server has been fixed
- RTCP scheduling improvements and reduced size RTCP support
- MP4/MOV muxer acquired a new “robust” mode of operation which attempts to keep the output file in a valid state at all times
- Live mixing support in aggregator, audiomixer and compositor was improved a lot
- compositor now also supports rescaling of inputs streams on the fly
- New audiointerleave element with proper input synchronisation and live input support
- Blackmagic Design DeckLink capture and playback card support was rewritten from scratch; 2k/4k support; mode sensing
- KLV metadata support in RTP and MPEG-TS
- H.265 video encoder (x265), decoders (libav, libde265) and RTP payloader and depayloaders
- New DTLS plugin and SRTP/DTLS support
- OpenGL3 support, multiple contexts and context propagation, 3D video, transfer/conversion separation, subtitle blending
- New OpenGL-based QML video sink, Gtk GL video sink, CoreAnimation CAOpenGLLayerSink video sink
- gst-libav switched to ffmpeg as libav-provider, gains support for 3D/multiview video, trick modes, and the CAVS codec
- GstHarness API for unit tests
- gst-editing-services got a completely new ges-launch-1.0 interface, improved mixing support and integration into gst-validate
- gnonlin has been deprecated in favor of nle (Non Linear Engine) in gst-editing-services
- gst-validate has a new plugin system, an extensive default testsuite, support for concurrent test runs and valgrind support
- cerbero build tool for SDK binary packages gains new ‘bundle-source’ command
- Various improvements to the Android, iOS, OS X and Windows platform support
修复的 bug 有:
- Crashes in the gst-libav encoders were fixed
- More DASH-IF test streams are working now
- Live DASH, HLS and MS SmoothStreaming streams work more reliable and other fixes for the adaptive streaming protocols
- Reverse playback works with scaletempo to keep the audio pitch
- Correct stream-time is reported for negative applied_rate
- SRTP packet validation during decoding does not reject valid packets anymore
- Fixes for audioaggregator and aggregator to start producing output at the right time, and e.g. not outputting lots of silence in the beginning
- gst-libav’s internal ffmpeg snapshot was updated to 2.8.1
- cerbero has support for Mac OS X 10.11 (El Capitan)
- Various memory leaks were fixed, including major leaks in playbin, playsink and decodebin
- Various GObject-Introspection annotation fixes for bindings
- and many, many more