The SVT-AV1 4.2.0 (Scalable Video Technology AV1) library has been released, featuring implementations of the AV1 video encoding format encoder and decoder, accelerated by hardware parallel computing extensions present in x86_64 and ARM processors. The project was created by Intel in partnership with Netflix to achieve a performance level suitable for on-the-fly video transcoding and use in video-on-demand (VOD) services. The development is currently underway under the auspices of the Open Media Alliance (AOMedia), which oversees the development of the AV1 video encoding format. The project was previously developed within the OpenVisualCloud project, which also develops the SVT-HEVC and SVT-VP9 encoders. The code is distributed under a BSD license.
SVT-AV1 can be compiled for systems based on any architecture with a compiler supporting the C99 standard, but best performance is achieved on x86_64 systems, which utilize assembler optimizations based on SIMD instructions (AVX2 support is preferred in the CPU, but SSE2 is sufficient as a minimum). Memory consumption depends on the number of processor cores used for encoding, which is controlled by the "--lp" option. Due to the complexity of the algorithms used in AV1, encoding this format requires significantly more resources than other formats.
Among the changes in the new SVT-AV1 release:
- A new TUNE-VMAF optimization mode has been implemented, allowing for bitrate savings of up to 15% while maintaining the same level of visual quality according to the VMAF metric, which evaluates human perception, and a slight reduction in quality according to the PSNR metric, which evaluates the accuracy of reproduction of the original data at the pixel level.
- A single-threaded encoding mode has been implemented, which can be used with the Random Access mode.
- In presets M3-M5, the quality/speed tradeoffs when encoding in random access mode have been optimized.
- Added new command line options: --cqp, --enable-intrabc, --hbd-mds, --enable-kf-tf.
- Added support for output in raw OBU format in addition to the previously supported IVF format.
- The quality/speed tradeoffs when encoding in "—rtc" (Real-Time Communications) mode have been optimized, compression efficiency has been increased, and memory consumption has been reduced when encoding at low resolutions. The "—max-intra-bitrate-pct" and "—max-inter-bitrate-pct" parameters have been added.
- A real-time constant bitrate encoding mode has been added—CBR (Constant Bitrate), implemented using a Kalman filter and supporting cyclical updating with the smearing of complex changes over several frames and re-encoding of frames with worse quality if they do not fit within the specified bitrate.
- In the "--rtc" mode, support for dynamically changing bitrate settings and presets on the fly has been added.
- Added API for managing keyframes, supporting LTR (Long-Term Reference) and two-level structure of reference picture sets (RPS).
- The implementation of entropy coding has been redesigned to provide direct writing to the tile-buffer, simplify the arithmetic coder, and implement trimming of residual data after quantization.
- Added new optimizations: CDEF (Constrained Directional Enhancement Filter) to eliminate blur in areas of contrast change, MD (Mode Decision) and ME (Motion Estimation) for fast processing of static objects.
- Optimized detection of the presence of an image with screen content in a video to apply specific encoding methods to preserve text clarity.
- Added build option ENABLE_STACK_PROTECTOR to disable the "-fstack-protector" flag, which removes the overhead of stack overflow checking.
- Added additional optimizations for the ARM architecture and leveraged extended NEON instructions to speed up computations.
Source: opennet.ru
