After seven months of development, the FFmpeg 8.1 multimedia package has been released, which includes a set of applications and a collection of libraries for operations with various multimedia formats (recording, converting, and decoding audio and video formats). The package is written in C and is distributed under the LGPL and GPL licenses.
Among the changes in FFmpeg 8.1:
- A parser, encoder, decoder, as well as a muxer and demuxer for the JPEG XS image format have been added. JPEG XS is positioned as a lightweight image coding system designed for minimal latency during encoding and decoding, aiming to optimize the transmission of sequences of very high-quality images (up to 8K). JPEG XS significantly reduces the bandwidth required without noticeable quality loss for the human eye. The implementation is based on the libsvtjpegxs library.
- An experimental decoder for the xHE-AAC (High-Efficiency Advanced Audio Coding) audio encoding format with Mps212 (MPEG Surround with 212 channel layout) has been added. xHE-AAC is used in Netflix streaming and involved in Digital Radio Mondiale technologies. The codec is notable for its support of a wide range of bitrates (from 12 to 300 kbit/s), high compression efficiency, playback at consistent volume, high clarity at any volume levels, additional dynamic range control profiles for listening in noisy environments, and metadata that allows recovery from losses on the receiving side.
- Based on the libmpeghdec library, a decoder for interactive and spatial audio in the NGA (Next Generation Audio) format, defined in the MPEG-H audio and video coding standard, has been implemented.
- Support for packing and unpacking spatial audio in the IAMF (Immersive Audio Model and Formats) format, featuring Ambisonics spatial sound that takes into account sound propagation not only in the horizontal plane but also vertically (to determine whether the sound source is above or below), has been added.
- Support has been added for parsing and redirecting metadata in the LCEVC (Low Complexity Enhancement Video Coding) format, which implements an additional layer of metadata on top of standard codecs to improve video quality. Support for exporting LCEVC quality enhancement layers into MPEG-TS (MPEG Transport Stream) multimedia containers has also been added.
- Based on the Vulkan graphics API, an encoder and decoder for the Apple ProRes codec have been implemented, as well as a decoder for the DPX (Digital Picture Exchange) format used in film production. The Vulkan-based implementations are notable for significantly improved performance due to hardware acceleration, parallel operations, and the use of compute shaders. Optimization has been made for AI Vulkan-based codec implementations. For faster codec initialization, the capability to use already compiled GLSL shaders without needing to compile them during operation has been implemented.
- The swscale (Software Scaler) library used in FFmpeg for software scaling and color conversion has implemented a backend that utilizes the Vulkan graphics API to accelerate operation performance.
- New encoder options for H.264 and AV1 formats using the D3D12 (Direct3D 12) API for hardware-accelerated encoding have been added.
- A variant of the H.264/HEVC encoder has been added that uses the hardware video encoding capabilities available in Rockchip chips.
- Demuxers for multimedia containers in HXVS and HXVT formats used in IP cameras have been added.
- A parser for EXIF metadata and a corresponding API for parsing metadata have been implemented.
- The ffprobe utility has added the option "-codec" ("-c") for selecting a specific decoder implementation.
- The ffmpeg utility has added support for tiled image storage in HEIF format (where a very large image is stored as a set of smaller images).
- The old HLS protocol handler has been removed.
- New filters:
- drawvg for outputting vector graphics over video frames using the libcairo library.
- vpp_amf for resizing video and converting color space using AMD Advanced Media Framework for hardware acceleration.
- vf_scale_d3d12, vf_deinterlace_d3d12, vf_mestimate_d3d12 for scaling, deinterlacing, and motion analysis on video, utilizing the Direct3D 12 graphics API for hardware acceleration.
- gfxcapture for capturing window and screen content on the Windows platform using the Windows.Graphics.Capture API.
- A bitstream filter for LCEVC metadata has been added.
Source: opennet.ru
