Slicer release 22022301
Fixes:
-
Fix CameraSlicer startup race condition.
-
Fix unreadable CameraSlicer UI in Dark mode (opt-out Slicer and CameraSlicer app from Dark mode).
-
Fix automatic gain adjustment of Dolby subtracks in VOD slicer.
-
Fix TTML-to-CC conversion when TTML contains non-Latin text.
-
Fix an issue where a slice was always truncated by up to 0.128s due to a short audio track.
-
Fix VOD slicer ingest of assets where some audio tracks are short.
-
Fix slicebot errors on CentOS 7.
-
Fix slicer log spam due to UL_EXPECT failures.
-
Fix video-only VOD ingest regression that was introduced in 21102800-master slicer release.
Enables:
-
Signal record/replay functionality for AJA liveslicer.
-
New prefetch_pod APIs for Ad prefetch workflows.
-
Optional target_loudness slicer config and command line parameter.
-
Optional dynamic gain adjustment via either -enable_loudnorm parameter (adds 3s latency, computationally expensive), or -enable_dyn_gain which adds 100ms latency and uses ebur128 output loudness measurements and incremental gain changes to reach -target_loudness.
-
Support for non-standard number of audio channels (up to 30).
Other changes:
-
Make render_608 buffer match the slice duration, by default.
-
Update signal record/replay functionality for BlackMagic, CDI, and UDP liveslicer (trailing_capture only), add support for overlapping captures, make output file path configurable.
-
Reimplement gain adjustment.
-
Change x264 encoder timebase to match target framerate.
-
Include Advanced Audio Routing settings (-mix ... -mix end) during VOD slicer analysis pass.
Prerequisites:
-
Due to the slicer being built on CentOS 7 it picks up a new library dependency on libnl-3.200, which is not included in the base install of Ubuntu 18.04. You may need to pre-install it: sudo apt install libnl-3-200
-
Nvidia driver 418 or later is required for HEVC 4K HDR/SDR.
-
Due to CentOS 6 reaching EOL in November 2020 this slicer release requires CentOS 7 or newer.