Samsung and Prime Video launch HDR10+ ADVANCED streaming format

This morning, Samsung and Amazon Prime Video quietly dropped a specification that could reshape how streaming platforms handle high Dynamic range content across the flood of heterogeneous display environments. They call it HDR10+ ADVANCED. And it's not just a minor bump to the royalty‑free HDR10+ standard - it adds a system for ambient light aware, scene‑by‑scene tone mapping that works without user calibration. The new HDR10+ ADVANCED format from Samsung and Prime Video introduces ambient light‑aware dynamic metadata, promising a unified display adaptation layer - but is it enough to unseat Dolby Vision? For engineering teams building streaming infrastructure, mobile playback SDKs. Or encoding pipelines, this move has deep technical implications that go far beyond a press release.

Most of us have seen the HDR spec wars play out in code. On any given Android device, you might get HDR10, or HDR10+. Or Dolby Vision - and that's if the OEM's MediaCodec implementation even bothered to expose the right profiles. Adding another format, even an extension, means more metadata parsing, more on‑the‑fly negotiation. And a whole new integration surface for QA automation. But it also means forcing a rethink of how we handle adaptive brightness in playback stacks. Which is something we've been hacking around with AmbientDisplay APIs and custom shaders for years.

In this piece, I'll walk through what HDR10+ ADVANCED actually adds to the existing standard, why the engineering challenge of ambient tone mapping is thornier than most assume and what it means for mobile developers, encoding teams. And anyone who's ever had to explain to a product manager why the same HDR stream looks different on two TVs in the same room. I'll pull from real integration experiences with ExoPlayer's HDR handling, AWS Elemental MediaConvert's HDR10+ packaging. And the fundamental SMPTE ST 2094 specification that underpins it all.

The Evolution of HDR Standards in Streaming: HDR10, HDR10+, and Dolby Vision

To understand why HDR10+ ADVANCED matters, you need to appreciate the landscape it's entering. The baseline HDR10 format (SMPTE ST 2086) provides static metadata: the entire feature gets one set of color volume parameters - max frame‑average light level, MaxCLL, MaxFALL - applied uniformly. This works, but it's a blunt instrument. Dolby Vision cracked this early with dynamic metadata, allowing brightness and tone mapping to shift per scene. Or even per frame, using the concept of an "artistic trims" pass that's part of a creative intent layer. That proprietary system, built on SMPTE ST 2094‑10, carries a licensing burden.

Samsung, with Panasonic and later 20th Century Fox, introduced HDR10+ as a royalty‑free alternative that also uses dynamic metadata. But anchored to SMPTE ST 2094‑40 (a different application identifier). Instead of a Dolby‑controlled composition, HDR10+ metadata - inserted as SEI messages in the HEVC bitstream - describes a Bezier‑curve‑based tone mapping function that a display can apply adaptively. Both HDR10+ and Dolby Vision solve the "static metadata" limbo, but they diverge significantly in the ecosystem, SDK support. And test complexity. In production, we've seen HDR10+ adoption primarily in Samsung TVs, a selection of Panasonic and Hisense panels and Prime Video's catalog, whereas Dolby Vision enjoys broader manufacturer support and a deeper creative toolchain.

Now enters HDR10+ ADVANCED. It layers additional logic on top of the ST 2094‑40 foundation to account for ambient light conditions in real time, effectively extending the metadata model to include display environment awareness - something that previously required separate machine‑learning tone mapping or manual presets. For engineers, this means the spec documents now likely define new metadata payloads or additional table identifiers that a renderer must consume alongside the existing scene‑by‑scene curves. Early developer-facing material from Samsung hints at a new "Advanced Tone Map" mode that can be signaled per frame.

What Exactly Is HDR10+ ADVANCED? Technical Overview of the New Ambient Metadata Layer

Based on the announcement details and supporting materials from HDR10+ Technologies, LLC, the ADVANCED extension introduces a standardized way for displays to perform automatic, light‑level‑aware rendering without relying on proprietary post‑processing. Under the hood, this is likely accomplished by adding a small set of ambient adaptation parameters to the dynamic metadata SEI payload. While HDR10+ already defines a mapping curve anchored to a reference peak brightness (e g., 1000 nits), HDR10+ ADVANCED could allow an originating content device - a mastering monitor or encoding pipeline - to specify alternative curve presets for low, medium. And high ambient light levels. The player's compositor then blends between these curves based on real‑time sensor readings.

Consider what this means for a mobile phone playing back a Prime Video title in a sunny park. Without ADVANCED, the app might either display a dim, underexposed image because the display can't hit 1000 nits in direct sunlight. Or it might rely on a heuristic brightness boost that easily crushes highlights. With ADVANCED's metadata, the playback stack can interpolate into a pre‑defined "bright environment" curve that preserves shadow detail while mapping the peak luminance to whatever the screen can actually achieve under that ambient light. This isn't a trivial conversion: tone mapping functions are invertible only under certain constraints. And any interpolation between curves must respect the non‑linear nature of the PQ (ST 2084) transfer function. A slight miscalculation at the knee of the curve and specular highlights blow out or midtones turn muddy.

For software teams, this will mean new capabilities to handle in the parse‑then‑render pipeline. On Android - for example, the MediaFormat key for HDR10+ metadata (currently KEY_HDR10_PLUS_STATIC_INFO) may be extended, or a new buffer flag will mark frames that carry ADVANCED‑specific data. The ExoPlayer team will almost certainly need to update their Hdr10PlusMetadata class to parse additional fields. Our internal test harness already uses FFmpeg's bitstream filter for HDR10+ extraction; ADVANCED will require upstream patches to demux those new SEI NAL unit types. Anyone who's wrestled with AWS Elemental MediaConvert's HDR10+ passthrough knows that even a minor spec change can break the global dependency chain.

The Role of Dynamic Metadata in Modern Rendering Pipelines

To see why ADVANCED is a logical progression, let's dissect how dynamic metadata currently flows from encoder to glass. In a typical HDR10+ encode, the content is mastered on a reference monitor calibrated to a known peak luminance. During the grading process, metadata authoring tools (like Samsung's own HDR10+ studio plugins) generate a series of tone mapping control parameters - anchor points for a spline - trim values, target display peak luminescence - per scene cut. These get packaged as ITU‑T T. 35 metadata with application identifier 4, embedded into HEVC supplemental enhancement information messages. The decoder side extracts these SEI messages, passes them to a compositor. And the compositor uses them to remap the HDR content from the reference mastering display to the actual display's capabilities.

In software stacks, the decomposition looks something like: elementary stream → MediaExtractor → MediaCodec → HDR metadata buffer (flag FLAG_HDR_INFO) → Surface with HDR10+ capability → GPU tone mapping shader. ExoPlayer's Hdr10PlusMetadata stores targetedSystemDisplayMaximumLuminance, averageMaxRgb. And the distributionMaxRgbPercentages that define the Bezier curve. The renderer then passes these into a VkPhysicalDevice or OpenGL ES shader that applies the transformation. The entire pipeline assumes a single reference viewing environment. HDR10+ ADVANCED will need to multiplex multiple environment‑conditioned curves into the same metadata flow. And the renderer will then need to choose or interpolate based on data from the Ambient Light Sensor API (Android's Sensor. TYPE_LIGHT).

This introduces non‑trivial statefulness into the rendering loop. When a user walks from a dark room into sunlight, the ambient light change triggers a sensor event; the playback stack must force a redraw of the current frame with a different tone map, ideally with hysteresis to prevent flickering. This is territory that previously lived in television‑side firmware (e, and g, Samsung's "Adaptive Picture" mode). Moving that decision into the app layer - where it can interact with streamed metadata from the cloud - is a profound architectural shift.

Encoding and Packaging HDR10+ ADVANCED Content for Prime Video

The announcement emphasized Prime Video's role as the launch distribution partner. Which makes sense: Amazon already has a mature HDR10+ encoding pipeline through AWS Elemental MediaConvert and a large catalog that's been mastered in HDR. For any engineering team responsible for delivering HDR10+ content, the first question is: what changes in the job config? Currently, a MediaConvert job with HDR10+ can be turned on with a simple checkbox in the output settings, provided the source mezzanine includes the HDR10+ metadata. ADVANCED likely introduces a new encode mode - something like "HDR10+ Advanced Tone Mapping" - that instructs the encoder to generate additional metadata tables for multiple ambient states.

This will almost certainly require new metadata authoring tools in the pre‑processing stage. Colorists will need to define not just a single master grade, but a set of "ambient trims" for at least three conditions: dark room (30 lux), normal (100 lux), and bright (300+ lux). The mastering suite then emits these as separate trims within the same metadata stream. From a file packaging perspective, this stays as SEI messages. But now there might be multiple metadata blocks per video frame. Heuristic parsers that treat HDR10+ metadata as a single contiguous array per decoded picture will need to handle multiple instances and order them correctly by ambient condition tag.

One open question is whether ADVANCED metadata will be backward compatible - can an existing HDR10+ decoder ignore the ambient‑specific sub‑structures and still render a standard HDR10+ image? The HDR10+ consortium has historically prioritized backward compatibility, so it's plausible. If so, packaging teams can adopt a progressive enhancement strategy: encode with ADVANCED. And older players simply drop the new data and render as before. But as we learned during the initial HDR10+ rollout, assumptions about MediaCodec's handling of unknown SEI payloads can be risky. We'll need extensive device testing to guarantee that the presence of new SEI types doesn't trigger parse errors or decoder hangs, especially on low‑end chipsets.

Device‑Side Adaptation: Why Ambient Light Sensing Changes the Game for Mobile Engineering