Overview
ISO/IEC 14496-15:2024/Amd 1:2025 is an amendment to the international standard for the carriage of network abstraction layer (NAL) unit structured video in the ISO base media file format. Developed by ISO and IEC, this amendment enhances the existing framework by introducing support for neural-network post-filter supplemental enhancement information (SEI) and refining mechanisms for signaling and processing SEI messages in video bitstreams. These updates improve how advanced video coding technologies, such as AVC, HEVC, and VVC, manage and deliver enhanced video experiences, particularly by incorporating neural network-driven post-processing and additional metadata signaling.
Key Topics
-
Supplemental Enhancement Information (SEI) Improvements
The amendment refines the organization and signaling of SEI messages, allowing better distinction between required and optional metadata. The SEI Information Box structure clarifies which SEI messages are essential for correct playback and which can provide supplementary enhancements.
-
Neural-Network Post-Filter Support
It introduces detailed specification for carrying information about neural-network post-filters (NNPF) via dedicated SEI messages (NNPFC and NNPFA) and associated sample groups.
- NNPFC SEI messages define characteristics or updates of neural network post-filters.
- NNPFA SEI messages signal activation of specific post-filters during playback.
-
Sample Groups and Metadata Integration
The standard defines new sample groups and item properties for managing neural-network post-filter information, facilitating efficient organization and access during media playback. This includes mechanisms to handle base filters and incremental updates.
-
Restricted Video Tracks and SEI Processing Requirements
Updates strengthen how file authors can mandate player or renderer actions based on SEI message presence and type, using a restricted video track scheme and clear signaling of essential SEI payloads.
Applications
-
Enhanced Video Playback and Quality
By supporting neural-network post-filter metadata, the amendment enables advanced post-processing such as denoising, upscaling, and content enhancement on compatible video players. These neural network models can be specified, updated, and activated dynamically within the video bitstream, improving visual quality adaptively.
-
Professional Broadcast and Streaming
Broadcasters, OTT platforms, and video streaming services can use these new capabilities to deliver higher-quality video to end-users by leveraging state-of-the-art neural network enhancements in a standardized, interoperable fashion.
-
Rich Metadata and Interoperability
The clarified handling of SEI messages ensures that critical metadata needed for accurate playback and rendering is effectively signaled, contributing to robust interoperability across encoders, packaging tools, and decoders supporting the ISO base media file format.
Related Standards
These related standards define the underlying video codecs, as well as the specific supplemental enhancement information structures referenced in the amendment.
Keywords: ISO base media file format, network abstraction layer, NAL unit, neural-network post-filter, SEI message, video coding, AVC, HEVC, VVC, metadata, sample group, video enhancement, ISO/IEC 14496-15.