LCEVC
Low Complexity Enhancement Video Coding (LCEVC) is a video compression standard developed by the Moving Picture Experts Group (MPEG) as part of the MPEG-5 suite, formally published as ISO/IEC 23094-2:2021 (with Amendment 1 in 2024 adding additional levels), that adds a lightweight enhancement layer to existing base video codecs such as H.264/AVC, HEVC, or AV1 to improve compression efficiency and video quality while maintaining low computational complexity.[1][2][3] The development of LCEVC was initiated in October 2018 in response to an industry call for proposals from 28 leading organizations in streaming, virtual reality, and broadcasting, aiming to create a tool that bridges the gap between legacy codecs and more advanced ones without requiring significant hardware upgrades.[1][4] After rigorous testing and validation during MPEG's collaborative phase, the standard was approved and published by the International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC) in November 2021.[5][6] Technically, LCEVC operates by encoding a lower-resolution base layer using a conventional codec and then applying one or more enhancement sub-layers that correct residuals—the differences between the original video and the base layer—through specialized, low-overhead tools such as 2x2 or 4x4 transforms, adaptive filtering, and upscaling filters, without relying on inter-block prediction to ensure parallel processing and minimal memory usage.[7] This design makes LCEVC codec-agnostic, allowing it to integrate seamlessly with a wide range of base technologies, and supports scalable delivery formats like MPEG-DASH and HLS for adaptive streaming.[1][8] LCEVC delivers notable benefits, including up to 40% bitrate savings or equivalent quality improvements compared to base codecs alone, with encoding times around 30-50% of base codecs and decoding overhead limited to about 10% additional compute resources, enabling efficient deployment on legacy devices, mobile platforms, and web browsers via software implementations.[7][8] These advantages have facilitated its adoption in broadcast, over-the-top streaming, and emerging applications like 4K/8K video and virtual reality, including selection for Brazil's TV 3.0 standard (with rollout as of 2025) and integrations in commercial encoders and players from companies such as V-Nova and MainConcept.[9][10][11]Introduction
Definition and Scope
LCEVC, or Low Complexity Enhancement Video Coding, is formally designated as MPEG-5 Part 2 and standardized under ISO/IEC 23094-2:2021 by the International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC). Developed by the Moving Picture Experts Group (MPEG), it represents a video coding standard specifically engineered to provide low-complexity enhancements to existing video compression technologies, enabling improved video quality and efficiency without requiring substantial increases in computational resources. This standard was finalized in November 2021 after development initiated in October 2018.[12] The scope of LCEVC is narrowly focused on serving as an additive layer to a pre-existing base video codec, rather than functioning as a comprehensive compression solution on its own. It incorporates one or two enhancement layers that address residuals—differences between the base layer reconstruction and the original video—and up-sampling for higher resolutions, all while maintaining low decoding and encoding complexity to suit resource-constrained environments such as mobile devices and live streaming applications.[13] This design ensures that LCEVC operates atop a base layer encoded using another established codec, thereby extending rather than replacing foundational compression mechanisms. In contrast to standalone full codecs such as AV1 or Versatile Video Coding (VVC), which handle complete video compression from scratch, LCEVC mandates the use of a base codec like H.264/AVC, High Efficiency Video Coding (HEVC), or AV1 to generate the initial layer, concentrating exclusively on enhancement functionalities to refine output quality and bitrate efficiency. This dependency distinguishes LCEVC as a complementary tool, promoting interoperability with legacy and modern video infrastructures without necessitating full system overhauls.[12]Purpose and Benefits
LCEVC, or Low Complexity Enhancement Video Coding, was developed primarily to enable the delivery of next-generation video compression efficiency, such as HEVC-level performance, on legacy hardware and software environments by significantly reducing decoding complexity compared to standalone advanced codecs. By layering enhancements atop existing base codecs like AVC or HEVC, LCEVC allows devices without dedicated hardware acceleration for newer standards to achieve comparable quality while lowering the computational demands of decoding by up to 50%, thereby extending battery life and improving playback efficiency on mobile and edge devices. Notable adoptions include its selection for Brazil's TV 3.0 standard in December 2021 and hardware integrations as of 2025, enhancing broadcast and streaming efficiency.[14][15] This approach addresses key challenges in video streaming and broadcasting, where computational constraints often limit the adoption of more efficient but complex codecs. A core benefit of LCEVC lies in its backward compatibility with established base codecs, ensuring seamless integration into existing workflows without requiring full codec replacements or hardware upgrades. This software-upgradable nature facilitates deployment on a wide range of devices, from smartphones to set-top boxes, supporting resolutions up to 8K while maintaining low overhead. In live streaming scenarios, LCEVC enhances sharpness and detail preservation by reconstructing high-frequency elements efficiently, with minimal added latency—often under 400 ms—making it suitable for real-time applications like interactive video and broadcasting.[16][17][18] In terms of efficiency, LCEVC delivers bitrate savings of 30-50% when enhancing H.264 (AVC) streams to achieve HEVC-equivalent quality, as verified in MPEG tests showing approximately 46% savings at UHD resolution and 28% at HD. These gains stem from encoding a lower-resolution base layer and applying targeted enhancements, reducing overall bitrate without compromising visual fidelity, particularly beneficial for bandwidth-constrained environments.[19]Technical Overview
Encoding Mechanism
The LCEVC encoding process begins with downsampling the input video to a lower resolution by a factor of 1.5 or 2 in each dimension using 2D scaling, or horizontal-only 1D scaling, or no scaling, with methods like bilinear or Lanczos filtering to preserve essential details. This downsampled video is then encoded using a legacy base codec, such as AVC/H.264 or HEVC/H.265, at reduced quality or resolution to form the base layer, which ensures compatibility with existing hardware and workflows. Residuals are subsequently generated by upsampling the base layer reconstruction and computing the difference with the downscaled input; these residuals capture high-frequency details, edges, and artifacts not preserved in the base layer. The residuals are encoded using lightweight tools, including small-kernel Discrete Cosine Transform (DCT) operations (2×2 or 4×4 blocks) for spatial frequency decomposition, followed by quantization and adaptive filtering to smooth block boundaries and reduce artifacts. LCEVC supports 2D scaling, 1D horizontal scaling, or no scaling for the base layer.[7][13] LCEVC supports up to two enhancement layers to progressively refine the base layer. The first enhancement sub-layer (L-1) encodes residuals between the downscaled input and the base layer reconstruction to correct compression artifacts at the base resolution. These residuals are sparse and primarily encode structural differences. The second enhancement sub-layer (L-2) addresses quality residuals, subtracting the upsampled L-1 reconstruction from the full-resolution original to capture fine details and texture; it incorporates low-complexity prediction modes, including intra-frame prediction for spatial redundancy and optional inter-frame temporal prediction using zero-motion vectors for block-based matching across frames. Entropy coding for both layers simplifies inherited base codec methods, utilizing Run-Length Encoding (RLE) for sparse zero runs and prefix codes for coefficient amplitudes, which minimizes overhead while maintaining efficiency.[7][20] This hybrid approach results in overall encoding complexity comparable to or lower than the base codec at full resolution, due to the lower-resolution base encoding and efficient residual tools, enabling seamless integration into broadcaster hybrid workflows without substantial computational demands.[7][13]Decoding Process
The decoding process of LCEVC reconstructs the enhanced video by first decoding the base layer bitstream using a standard video decoder, such as H.264/AVC or H.265/HEVC, which utilizes existing hardware and software implementations for efficiency.[7] The resulting decoded base picture, typically at reduced resolution, is then up-sampled to the target resolution using lightweight filters such as nearest-neighbor, bilinear, cubic, or adaptive cubic kernels to generate a preliminary intermediate picture.[21] These residuals, derived during encoding from differences between the original video and the downscaled base layer, form the basis for subsequent enhancements.[7] The first enhancement sub-layer (L-1) is subsequently decoded from the enhancement bitstream using simple inverse quantization and inverse transform operations on small 2×2 or 4×4 kernels, then added arithmetically to the up-sampled base picture to form a combined intermediate picture, primarily improving resolution and correcting base-layer impairments.[7] If further up-sampling is required, it is applied to this combined picture to create a preliminary output picture at full resolution.[21] The second enhancement sub-layer (L-2) is then decoded, optionally incorporating temporal prediction from reference frames, and added via basic arithmetic to the preliminary output, yielding the final combined output picture with enhanced quality and detail.[7] LCEVC's decoding pipeline emphasizes low complexity by avoiding the need for specialized hardware, instead leveraging general-purpose CPU or GPU resources for the enhancement stages, which supports software-based implementations and fallback to base-layer rendering on legacy devices.[7] The overall process incurs minimal overhead, with total decoding complexity comparable to the base codec at full resolution, or adding approximately 10-45% overhead depending on the base codec and hardware, facilitated by parallelizable operations and sparse residual processing.[20] Following enhancement addition, adaptive de-ringing and noise reduction filters are applied to the reconstructed picture to minimize artifacts such as ringing around edges and banding, distinctive to LCEVC's efficient design; these include the L-1 smoothing filter for boundary artifacts and signaled dithering for noise suppression.[13]Development and Standardization
Historical Timeline
In 2018, the video compression industry identified a need for low-complexity enhancement technologies to extend the capabilities of existing codecs, driven by demands for improved efficiency in bandwidth-constrained environments and legacy hardware compatibility.[7] This led to the Moving Picture Experts Group (MPEG) issuing a Call for Proposals (CfP) in October 2018, soliciting technologies for a new standard focused on software-based enhancements to base video streams.[13] Responses to the CfP were evaluated by MPEG in early 2019, with five proposals undergoing initial testing for compression performance and complexity metrics.[7] Core experiments based on these evaluations were initiated later that year, incorporating contributions from industry leaders including V-Nova's proprietary P+ technology, which provided foundational bi-layer enhancement mechanisms that evolved into elements of the open standard.[22] Prototypes derived from these experiments were demonstrated at IBC 2019, showcasing real-time encoding and decoding improvements over base codecs like AVC and HEVC.[23] The core experiments progressed through iterative refinements, culminating in the finalization of key technical components by mid-2020, which informed the first draft of the standard.[7] In April 2021, during MPEG's 134th meeting, subjective quality validation tests confirmed the standard's performance, demonstrating significant bit-rate savings—such as approximately 30-50% when enhancing HD and UHD content—while maintaining low computational overhead. Early adoption emerged in December 2021, when the SBTVD Forum in Brazil selected LCEVC as the enhancement layer for its TV 3.0 standard, marking the first major broadcast integration following the standard's development.[14] This decision followed extensive testing and highlighted LCEVC's role in enabling next-generation television features like higher resolutions and immersive formats on existing infrastructure.[24] As of 2025, LCEVC supports Brazil's TV 3.0 rollout with live deployments, including enhanced 4K broadcasts, and has been demonstrated in ATSC 3.0 systems, with dedicated licensing programs launched in August 2025 to facilitate adoption in broadcasting and streaming.[25][26]Standardization Milestones
The formal standardization of LCEVC (ISO/IEC 23094-2) was managed by MPEG Working Group 04 (WG04), the subgroup focused on video coding technologies within the Moving Picture Experts Group (MPEG). The process involved multiple drafts, core experiments, and verification tests to ensure compliance with MPEG requirements, including maintaining encoding and decoding complexity at levels comparable to the base codec alone for full-resolution video processing.[13] The Final Draft International Standard (FDIS) was approved in October 2020 at the 132nd MPEG meeting, marking the completion of technical development and initiating the final approval ballot.[6] Following FDIS approval, LCEVC was published as the International Standard ISO/IEC 23094-2 in November 2021, defining the core encoding and decoding mechanisms for low-complexity enhancement layers compatible with base codecs such as AVC, HEVC, and VVC.[2] In 2023, extensions were introduced through amendments to related MPEG standards, including Amendment 1 to ISO/IEC 14496-15, which added support for carriage of LCEVC bitstreams and higher profiles to enable broader interoperability in file formats and transport streams.[27] Verification against MPEG requirements confirmed that LCEVC implementations achieve bit-rate savings while limiting complexity increases to under twice that of the base codec, as demonstrated in subjective tests conducted in May 2021.[13] As of 2025, LCEVC has been demonstrated in ATSC 3.0 ecosystems, supporting next-generation television systems with enhanced video delivery.[11] The core standard remains unchanged with no major revisions, though ongoing work focuses on refinements in supporting standards for emerging applications.[28]Licensing and Compatibility
Licensing Framework
The licensing framework for LCEVC (Low Complexity Enhancement Video Coding), standardized as MPEG-5 Part 2 (ISO/IEC 23094-2), is administered by V-Nova as the patent pool manager, encompassing essential patents contributed by multiple holders to enable fair, reasonable, and non-discriminatory (FRAND) terms.[29] Decoding capabilities and basic integration for encoding are provided free of charge to all users, including device manufacturers, chipset developers, operating systems, browsers, and in-house research and development efforts, facilitating widespread adoption without upfront costs for implementation.[30] However, commercial use of the enhancement layer, which provides the core efficiency gains when combined with a base codec, requires a license from V-Nova, with royalties structured on a volume-based model to align with deployment scale.[31] For video distribution services, such as streaming platforms, royalties are calculated monthly based on active user counts, subscriber numbers, or annual revenue, with tiered rates starting from a waiver threshold of up to 1.14 million active users, 114,000 subscribers, or $5.7 million in revenue, and scaling to a maximum of $667,000 per month for services exceeding 1.75 billion users, 175 million subscribers, or $8.75 billion in revenue; these fees are capped annually at $8 million for early adopters subscribing before December 31, 2026, and apply separately to each branded service.[31] Broadcast applications benefit from flexible terms, including full royalty waivers for free-to-air (FTA) transmissions, while paid broadcast services may opt for lump-sum arrangements tailored to operational needs.[32] Non-commercial and research uses incur no fees, as the free integration license explicitly covers development and testing activities without revenue generation.[30] Consumer device manufacturers, such as those producing televisions and set-top boxes, face per-unit royalties that decrease with volume—for instance, $0.70 per television unit for volumes up to 1 million, dropping to $0.35 for over 20 million units, with an annual cap of $20 million—covering the inclusion of LCEVC decoding hardware or software.[29] Additional incentives include discounted "open-access" rates if LCEVC is enabled across all media systems and codecs on the device without restrictions, promoting broader ecosystem compatibility. In August 2025, V-Nova launched dedicated licensing programs optimized for TV 3.0 and ATSC 3.0 deployments, providing streamlined terms to accelerate next-generation broadcast transitions in regions adopting these standards.[32]Compatibility with Base Codecs
LCEVC is designed to interoperate with a variety of established video codecs as its base layer, enabling enhancement without altering the underlying base codec infrastructure. It supports ISO-standard base codecs including H.264/AVC, HEVC/H.265, VVC/H.266, and EVC, as well as AV1, allowing it to function with any compliant base up to 8K resolution (8192×4320). This codec-agnostic approach ensures broad applicability across existing ecosystems, where the base layer handles primary compression and LCEVC adds low-complexity enhancements for improved quality or resolution.[13][33][34] The integration of the LCEVC enhancement stream with the base codec occurs through multiplexing, either by embedding the enhancement data as Supplemental Enhancement Information (SEI) messages within the base layer bitstream or by carrying it in separate tracks within standard containers such as MP4 or MPEG-2 Transport Stream (TS). This method preserves backward compatibility, as the base layer remains fully decodable by devices lacking LCEVC support, requiring only the presence of an unmodified base decoder for operation. No changes to the base codec's encoding or decoding processes are necessary, making LCEVC suitable for retrofitting legacy systems without hardware overhauls.[10][33][35] Particularly effective with older base codecs like H.264/AVC, LCEVC provides significant uplift for legacy deployments by enhancing compression efficiency and resolution on existing infrastructure. It employs up to two enhancement levels: level 1 focuses primarily on up-sampling the base layer to higher resolutions, while level 2 adds residual details for further quality improvements, all while maintaining low computational overhead.[36][14][13]Implementations and Adoption
Software Support
LCEVC has seen integration into several prominent open-source and commercial software libraries, enabling encoding and decoding capabilities across various workflows. FFmpeg, a widely used multimedia framework, introduced official support for LCEVC decoding in version 7.1 released in 2024, allowing developers to process LCEVC-enhanced streams within its command-line tools and libraries. This integration facilitates seamless enhancement of base codecs like H.264 and HEVC in video processing pipelines.[37] GStreamer, an open-source multimedia framework optimized for streaming, added a dedicated LCEVC plugin in version 1.26, released on March 11, 2025. This plugin includes both encoder and decoder elements based on V-Nova's SDK, supporting extraction of LCEVC enhancement data from H.264 streams and integration into real-time streaming pipelines for broadcast and OTT applications.[38] On the commercial side, MainConcept's SDK, updated in 2024, provides professional-grade LCEVC encoding tools compatible with its existing codec libraries for AVC, HEVC, and VVC base layers. This SDK integrates with Wowza Streaming Engine, enabling live encoding workflows that leverage LCEVC for bandwidth-efficient delivery over IP networks.[39] V-Nova offers reference software implementations of the LCEVC standard, serving as conformance tools for verifying encoder and decoder compliance with MPEG-5 Part 2 specifications. These reference tools, part of V-Nova's SDK, are utilized by developers and standards bodies to ensure interoperability and adherence to the ISO/IEC 23094-2 norm.[33] Open-source LCEVC decoders are available under the BSD-3-Clause-Clear license, including V-Nova's LCEVCdec library for native C++ decoding and LCEVCdecJS for web environments. The JavaScript-based decoder supports browser playback via HTML5, using WebAssembly to process LCEVC enhancements and render them to canvas elements, making it suitable for web streaming without native plugins.[40][41] In April 2025, MulticoreWare announced integration of V-Nova's LCEVC encoder libraries into its video processing suite, targeting broadcast workflows for enhanced codec efficiency. This partnership focuses on optimizing LCEVC for real-time encoding in TV 3.0 deployments, such as Brazil's digital TV transition, by layering enhancements atop x265 and x266 encoders.[42] In June 2025, Ateme integrated V-Nova's LCEVC SDK into its titra live encoding platform, demonstrating enhanced 4K HDR broadcast over ATSC 3.0 at the NextGen Broadcast Conference. This integration enables broadcasters to reduce bitrate requirements for UHD content while maintaining quality.[43]Hardware and Industry Integrations
LCEVC's hardware implementations leverage its design as an enhancement codec, allowing integration with existing base codec accelerators without requiring complete redesigns of video processing pipelines. Amlogic has incorporated native LCEVC decoding into its system-on-chips (SoCs) starting from 2024, enabling efficient enhancement processing directly in TV and streaming devices such as the SEI Robotics X5M Dongle launched in September 2024.[44][45] This approach utilizes hardware decoders for base layers like AVC, HEVC, or AV1, adding low-complexity enhancement stages that minimize additional silicon overhead.[9] V-Nova provides a development kit to facilitate custom silicon implementations, offering reference designs, ASIC IP blocks, and FPGA support for integrating LCEVC into new SoCs and video processors.[46] Released in 2022 and updated for ongoing deployments, the kit supports transitions from software prototypes to hardware-accelerated solutions, targeting broadcasters and device manufacturers seeking optimized performance.[46] A key advantage of LCEVC in hardware is its compatibility with software-based decoding on general-purpose CPUs within mobile SoCs, requiring no full hardware overhaul and resulting in only single-digit percentage increases in power consumption compared to base codec decoding alone.[47] Industry integrations have accelerated LCEVC's deployment through strategic partnerships. In April 2025, MulticoreWare and V-Nova collaborated to integrate LCEVC libraries into encoding platforms, supporting the Brazil TV 3.0 rollout with enhanced UHD delivery over broadcast networks.[42] This contributed to Globo's launch of experimental DTV+ pilots in Rio de Janeiro in April 2025, with set-top box vendors integrating LCEVC decoders for consumer trials. On August 27, 2025, Brazil's president signed a decree officially launching TV 3.0, enabling nationwide 4K HDR content delivery.[48][49] Similarly, MainConcept expanded LCEVC support in Wowza Streaming Engine in September 2024, enabling cloud-based encoding and delivery for low-latency streaming services.[39] These efforts extend to content delivery networks, where LCEVC enhancements are incorporated via SDKs for scalable video distribution in OTT and live applications.[50]Applications and Performance
Primary Use Cases
LCEVC finds primary application in live video streaming, where it enhances base layer codecs like AVC to deliver high-quality content such as 1080p60 video on mobile devices in bandwidth-constrained environments.[13] This approach is particularly suited for real-time encoding scenarios, including esports and live events, enabling low-latency transmission without requiring significant increases in computational resources.[13] In broadcast television, LCEVC facilitates upgrades to higher resolutions and formats like 4K and HDR by layering enhancements over existing base codecs, allowing delivery to legacy decoders without hardware replacements.[51] A notable example is Brazil's TV 3.0, launched in August 2025 with commercial deployment beginning in 2026, which leverages LCEVC-enhanced VVC base layers to enable 4K broadcasting on existing televisions nationwide.[51] For over-the-top (OTT) services, LCEVC enhances legacy content by improving visual quality and efficiency in adaptive bitrate streaming protocols such as DASH and HLS, making it ideal for mass-market distribution over variable network conditions.[1] It integrates seamlessly with digital rights management systems, supporting secure delivery in these environments.[52]Efficiency Gains and Comparisons
LCEVC delivers notable efficiency improvements by enhancing existing base codecs, achieving bitrate reductions of 40-60% compared to standalone AVC when targeting HEVC-equivalent quality levels, particularly for UHD content.[19] This stems from its layered approach, where a low-resolution base layer is supplemented by lightweight enhancement data, enabling higher visual fidelity at lower overall bitrates without requiring a full codec overhaul. For instance, formal MPEG verification tests demonstrated average bitrate savings of approximately 46% for UHD sequences when using AVC as the base, allowing LCEVC-enhanced streams to match or exceed HEVC performance while leveraging widely deployed AVC infrastructure.[19] In terms of computational demands, LCEVC maintains low overhead, with decoding complexity involving approximately 10% additional compute resources relative to the base codec alone, far below the 4-5 times increase observed for VVC relative to HEVC.[7] Encoding benefits are similarly pronounced in hybrid configurations, offering up to 3x speedup over traditional AVC encoding for equivalent quality, as the enhancement layer processes residuals efficiently without re-encoding the full frame.[53] These gains make LCEVC particularly suitable for resource-constrained environments, such as live streaming workflows, where reduced processing time directly translates to faster turnaround and lower operational costs. Compared to standalone next-generation codecs like AV1 and VVC, LCEVC exhibits lower overall complexity since it builds on established base codecs rather than operating independently, though it inherits the base's royalty and deployment requirements.[54] Against EVC, LCEVC offers broader compatibility with diverse base codecs (including AVC, HEVC, AV1, and even EVC itself) and incurs lower overhead for residual data compression, enabling more flexible integration across ecosystems.[7] Subjective quality assessments in MPEG validations further highlight LCEVC's advantages, with viewers showing a 20-30% preference for LCEVC-enhanced content over base codec outputs at matched bitrates, as measured by DSIS methodology and Mean Opinion Scores.[19]| Aspect | LCEVC + Base (e.g., AVC/AV1) | Standalone VVC | Standalone AV1 | EVC |
|---|---|---|---|---|
| Bitrate Efficiency | 40-60% reduction vs. base for HEVC-quality | 30-50% better than HEVC | 20-30% better than HEVC | 20-30% better than HEVC |
| Decoding Complexity | ~1.1x base (10% overhead) | 4-5x HEVC | 2-3x HEVC (software) | 1.5-2x HEVC |
| Encoding Speedup | Up to 3x vs. base | 5-10x slower than HEVC | 5-7x slower than HEVC | 2-3x slower than HEVC |
| Base Compatibility | Broad (AVC, HEVC, AV1, etc.) | None (standalone) | None (standalone) | Limited to EVC tools |
| Subjective Preference | 20-30% over base | Comparable to AV1 at high rates | Strong at low rates | 10-20% over HEVC |