After nine months of development multimedia package , including a set of applications and a collection of libraries for operations on various multimedia formats (recording, converting and decoding audio and video formats). The package is distributed under LGPL and GPL licenses, and FFmpeg development is carried out concurrently with the project .
From , In FFmpeg 4.2, the following features can be highlighted:
- Added the ability to use Clang for compiling CUDA files;
- Implemented support for decoding the AV1 format using an alternative decoder , developed by the VideoLAN and FFmpeg projects. Dav1d is focused on achieving maximum possible decoding performance and ensuring quality operation in multithreaded mode;
- Added media container demuxers for dhav, hcom, vividas, KUX, and IFV;
- Added PCM-DVD encoder;
- Added decoders for VP4, hymt, hcom, ARBC, agm, and lscr;
- Provided support for decoding HEVC 4:4:4 content using NVIDIA hardware acceleration engines nvdec and cuviddec, as well as using the VDPAU (Video Decode and Presentation) API;
- In the mov media container packager, tracks can now be recorded without explicitly defining the language (previously, English was the default);
- New :
- — automatic speech recognition using the PocketSphinx engine;
- — removes rain from video using a machine learning system based on neural networks; and ready ;
- — detects the absence of changes in the video (static image for a certain time);
- — adds additional frame sets to the beginning or end of the video stream;
- — smoothing brightness and color artifacts (spots and rainbows) in video;
- — shifts pixel color components horizontally and vertically;
- — extracts the base stream from TrueHD, discarding ATMOS metadata;
- — suppresses broadband noise in the audio stream using an algorithm ;
- — creates a mask based on the input video;
- — splits frames in the AV1 stream;
- — slows down the color change of dark pixels (increases the display time of bright flashes);
- — soft clipping of audio (gradually fading amplitude instead of abrupt signal cutoff);
- — removes information about all RGB colors except for the specified one;
- — extracts the median intersection of pixels from multiple input videos;
- — transforms stereo sound in the video, displaying the spatial interaction between two audio channels;
- — removal of distortions caused by poor quality microphones or excessive audio compression during voice recording (eliminates the canned sound effect).
- Includes a parser for GIF image format;
- Added support for subtitles in ARIB STD-B24 encoding (profiles A and C) used for MPEG-2 TS streaming. Support is implemented via the libaribb24 library;
- Removed the libndi-newtek library.
Source: opennet.ru
