NVIDIA has released the latest version of its Video Codec SDK, version 12.2, which significantly improves video quality, particularly for High Efficiency Video Coding (HEVC). According to the NVIDIA Technology Blog, the update provides a comprehensive range of APIs for hardware-accelerated video encoding and decoding on both Windows and Linux platforms.
Key features of Video Codec SDK 12.2
Video Codec SDK 12.2 introduces several new features aimed at improving video quality and reducing bitrates. These features include:
- Lookahead Level: This feature enhances predictive analytics to optimize coding efficiency, especially for HEVC. This allows for better bit allocation across frames.
- Temporal filtering: Designed to filter out noise and improve compression efficiency, this feature uses motion estimation to reduce the bitrate required to encode natural video content.
- Higher bit depth encoding: This feature allows you to encode 8-bit content to 10-bit to improve correlation and increase compression ratios for both HEVC and AV1.
- Unidirectional B-frame: These frames improve quality by only using predictions from past frames, making them useful for low-latency encoding scenarios.
- UHQ (Ultra High Quality) tuning information: This feature optimally combines predictive level and temporal filtering to provide the highest quality for latency-tolerant use cases.
Detailed feature insights
Preview level: This feature uses coding tree units (CTUs) and other encoding statistics to improve speed control. It offers four preview levels, each offering a different balance between performance and quality. Settings include: enableLookahead To turn on the feature: LookaheadDepth Determines the number of frames buffered for analysis. LookaheadLevel Set the level of analysis.
Temporal filtering: This feature uses patches from adjacent frames to reduce noise in natural video content. Temporal filtering, implemented using CUDA kernels for filtering and NVIDIA encoder (NVENC) hardware for motion estimation, provides an average coding gain of 4-5% for natural video content.
Higher bit depth encoding: By encoding 8-bit content to 10-bit, this feature provides a 3-4% coding efficiency improvement. The conversion is done within the NVENC driver using CUDA for HEVC and hardware for AV1, with virtually no performance penalty.
Unidirectional B-frame: This frame predicts the previous two frames, saving BD-BR by 3-4% compared to the P frame. This feature is especially useful for low-latency encoding and can be enabled through settings. enableUniDirectionalB To 1.
UHQ tuning information: This feature combines preview levels and temporal filtering to provide optimal settings for various presets, especially latency-tolerant encoding. We fix the number of B-frames to 5 and disable adaptive I and B-frames using a fixed GOP structure.
Getting started with Video Codec SDK 12.2
Video Codec SDK 12.2 significantly improves video encoding quality for HEVC with new features designed to reduce bitrate and improve overall performance. Users can download Video Codec SDK 12.2 to start and participate in discussions on the NVIDIA Developer Forum. For more information about new features, visit the NVIDIA Technology Blog.
Image source: Shutterstock