HDR Strategy: Scene-Referred vs Display-Referred Thinking

---

Fundamentals of Scene-Referred vs Display-Referred Signal Flows

At its core, the difference between scene-referred and display-referred signal flows lies in their reference point. A scene-referred signal aims to represent the actual light values present in the original scene, independent of any specific display device. Think of it as a photographic negative that holds all the captured information without committing to a particular print. This approach values the preservation of absolute light measurements and wide dynamic range. In contrast, a display-referred signal is optimized for a specific output device, taking into account its capabilities and limitations.

This is like tailoring a print for a particular type of paper and viewing conditions.

The established practice in professional HDR workflows starts in the scene-referred domain. Footage is treated as if it contains absolute light values, often in a linear light state or a logarithmic encoding designed to capture a vast dynamic range. Formats like OpenEXR, widely used in visual effects and computer graphics, exemplify this by storing multi-channel floating-point data that can represent an effectively unlimited dynamic range. This allows for precise manipulation of light values without loss of information, crucial for tasks like compositing or significant creative color adjustments.

Once creative adjustments are made in this flexible scene-referred space, the signal is then converted to a display-referred format. This conversion typically involves applying an Electro-Optical Transfer Function (EOTF), such as the Perceptual Quantizer (PQ) defined by SMPTE ST 2084. The PQ EOTF maps the scene-referred light values to a signal optimized for HDR displays, ensuring that the expanded brightness and contrast are accurately rendered. This process allocates digital values to specific light levels (cd/m², also known as nits) that a display can reproduce. For instance, HDR reference white is often set at 203 cd/m² (per ITU-R BT.2408), with peak highlights reaching 1000 cd/m² for a 14.3-stop dynamic range on many mastering monitors.

The benefit of this two-stage approach is flexibility. By operating in a scene-referred space for the bulk of creative work, filmmakers retain maximum control over the image's light and color. This allows for precise adjustments to exposure, contrast, and color without prematurely clipping highlights or crushing shadows. Only at the final stage, when preparing for a specific delivery target, is the image adapted to the display-referred domain. This adaptation might involve interpolation techniques, especially for handling the "headroom" (the brightness range above reference white) in HDR, which can be 1-4 stops depending on the content and display capabilities.

Tools like DaVinci Resolve Studio offer native support for working in Rec.2100 PQ, allowing for a smooth transition from scene-referred manipulation to display-referred output.

💡 Pro Tip: Node Labeling for Clarity

In node-based grading software, clearly label your nodes. Use labels like "Input Transform," "Primary Balance (Scene-Referred)," "Creative Looks," and "Output Transform (PQ)." This helps maintain awareness of where you are in the signal chain and prevents unintended display-referred manipulations too early in the process.

A common mistake is to grade directly into a display-referred space from the outset, often by applying a display LUT too early or by making creative decisions based solely on a non-HDR monitor. This can lead to grades that look good on a specific monitor but fall apart on other displays or when adapted to SDR. For example, the acclaimed Game of Thrones episode "The Long Night" received criticism for its dark, visually muddy appearance on many consumer SDR displays, a potential consequence of an HDR grade that didn't translate effectively to a lower dynamic range environment. Starting scene-referred gives you a solid foundation that can be flexibly adapted to various display targets, including SDR, ensuring better consistency across platforms.

For managing color across different stages, consider exploring concepts around Building a LUT Pipeline: Show LUTs, CDLs, and Governance.

Hardware Standards for Accurate HDR Monitoring

Accurate monitoring is paramount in HDR production. Without a reliable display, grading decisions are speculative, leading to unpredictable results on consumer screens. The fundamental requirement for any HDR reference monitor is its ability to accurately reproduce the expanded dynamic range and wider color gamut of HDR content.

Industry standards dictate strict performance criteria for HDR reference monitors. These displays must achieve a small-area peak white brightness of at least 1000 cd/m² and a black level of 0.05 cd/m², yielding a dynamic range of 14.3 stops. Critically, color accuracy is measured by Delta E, with values typically needing to be less than 2.3. Many high-end mastering monitors, like the Sony BVM-HX310, boast even lower Delta E values and deeper blacks, sometimes down to 0.0005 cd/m². These monitors also need to cover the Rec. BT.2020 color gamut, which encompasses approximately 75% of human vision, a significant expansion compared to Rec. BT.709's approximately 36%. The Eizo ColorEdge CG319X is another example of a monitor designed for HDR reference grading, offering 1000 cd/m² peak brightness and covering 99% of DCI-P3/BT.2020.

Calibration is not optional; it is mandatory. Hardware calibration with a professional colorimeter and accompanying software ensures that the monitor maintains its accuracy over time. For PQ mastering, the industry standard practice is to calibrate the reference white to 203 cd/m² and conduct grading in a dim surround environment of 5 cd/m². This controlled viewing condition minimizes the influence of ambient light on perception, allowing for consistent and repeatable grading decisions.

💡 Pro Tip: Dim Surround Matters

Always grade HDR in a controlled, dim environment with a neutral gray surround. The human visual system adapts to ambient light, and grading in a bright or color-biased room will lead to inaccurate perception of brightness and color, resulting in a grade that looks different (and likely worse) in a proper viewing environment. A 5 cd/m² dim surround is recommended for PQ mastering.

A common mistake filmmakers make is attempting to grade HDR on a non-HDR monitor or an HDR-capable monitor that hasn't been properly calibrated. An SDR monitor, typically limited to 100-300 cd/m², cannot display the expanded highlights and subtle shadow detail of HDR, and is therefore unsuitable for critical HDR grading, even if you are using it for rough checks. This leads to "flying blind" where highlights are clipped or crushed without the grader's knowledge, or shadows appear washed out on delivery. The visual information simply isn't there on the display. Even a consumer HDR TV, while capable of higher brightness, is rarely calibrated to reference standards and often applies its own processing, making it unsuitable for critical grading decisions.

For workflows involving both SDR and HDR deliverables, it is crucial to understand how displays adapt. Many professional monitors and software tools offer an Opto-Optical Transfer Function (OOTF) re-rendering capability. This allows for previewing how the HDR master will be adapted for consumer displays with different peak brightness capabilities, dynamically adjusting the reference white to match the display's characteristics. This is vital for ensuring that the creative intent holds up across a wide range of devices. See also our guide on on-set monitoring and video village setup for upstream display considerations.

Software Tools and Node-Based Workflows for HDR Grading

Modern color grading software, particularly DaVinci Resolve Studio and FilmLight's Baselight, forms the backbone of professional HDR workflows. These tools are designed to handle the complexities of high dynamic range and wide color gamut content, using node-based architectures to provide granular control over the image.

The standard practice involves working within a Rec.2100 PQ color space (or HLG, depending on the target), typically at 10-bit or 12-bit per channel, with the BT.2020 color gamut. The node-based workflow allows for a structured approach to grading. A common sequence might involve an input transform (LUT) to bring the source footage into the working color space, followed by primary balance and exposure adjustments. These initial steps are often performed in a scene-referred manner, manipulating the absolute light values.

Subsequent nodes are dedicated to creative adjustments, such as contrast, saturation, and specific color manipulations. For HDR, specialized tools like DaVinci Resolve's HDR palette provide intuitive controls for adjusting specific zones of brightness, from deep shadows to extreme highlights, without affecting other parts of the image. For instance, making a 1-stop adjustment in the HDR tab can have a profoundly different visual impact compared to an SDR adjustment, due to the expanded range. It is common to use parallel nodes for different creative adjustments, allowing for a non-destructive workflow where each adjustment can be independently tweaked or bypassed.

For integrating computer graphics, software like Nuke or Adobe After Effects handles 32-bit float EXR files, preserving the scene-referred data from the VFX pipeline, which can then be directly integrated into the grading suite.

💡 Pro Tip: Process Saturation Post-Contrast

When working in HDR, especially with wide gamut (BT.2020) material, process saturation adjustments after contrast adjustments. This helps prevent colors from becoming oversaturated or "breaking" in the extreme highlights or shadows. Using a color warper or specific hue-vs-saturation curves can also provide more nuanced control over specific colors without globally affecting the image.

A significant pitfall is oversaturating colors in HDR without using precise, hue-specific tools. The wider BT.2020 gamut means colors can be pushed further, but without careful control, they can appear unnatural or "neon." Tools like the color warper in Resolve allow for targeted adjustments to specific hues and their saturation/luminance, preventing global color shifts. Another common mistake is ignoring bit depth; grading in an 8-bit environment limits the color information to 16.7 million colors, whereas 10-bit HDR offers over a billion colors. This difference is visually apparent in gradients and subtle color shifts, where 8-bit can introduce banding.

Professional workflows often use the full capabilities of these software suites, including Dolby Vision mastering tools integrated into Resolve Studio. This allows for the creation of dynamic metadata which optimizes the HDR image for various consumer displays. Resolve's 32-bit float processing engine is crucial here, providing the necessary headroom and precision for complex HDR calculations and conversions. The ability to label nodes clearly (e.g., "input LUT," "temperature," "HDR") is a simple yet effective way to maintain order and understand the signal flow within a complex node tree, especially when revisiting a project or collaborating with others.

Practical Grading Techniques: Balancing Scene and Display Referral

The art of HDR grading lies in effectively balancing the creative intent, often conceived in a scene-referred mindset, with the technical requirements of display-referred output. The process typically begins with foundational adjustments that honor the scene-referred nature of the captured image, then gradually introduces display-referred considerations.

Initial grading passes focus on establishing a solid base. This involves primary color correction, white balance, and exposure adjustments, all of which should be performed in a scene-referred manner. Manipulating the linear or log encoded values ensures that you are adjusting the actual light captured, not merely remapping it for a specific display. For example, a 1-stop exposure adjustment in a scene-referred workflow directly corresponds to doubling or halving the light values, providing a predictable and physically accurate change.

Once the foundational balance is achieved, attention shifts to contrast and creative "looks." This is where HDR-specific tools become invaluable. Instead of a single master contrast control, HDR grading often involves more granular control over different brightness zones. Using curves or dedicated HDR palettes, you can manipulate highlights, mids, and shadows independently. For instance, gently rolling off extreme highlights to preserve detail or lifting shadows to reveal hidden information is more precise in HDR.

💡 Pro Tip: Separate Lights and Shadows Nodes

Always separate your light and shadow adjustments into distinct nodes early in your HDR grade. This allows you to make precise, independent adjustments to each region without inadvertently affecting the other. For example, you might have a node for "Shadow Lift" and another for "Highlight Roll-Off" before applying broader creative contrast.

The transition from scene-referred creative adjustments to display-referred output is managed by the EOTF, specifically the PQ curve for most HDR deliveries. However, this is not a one-size-fits-all solution. HDR content is consumed on a wide array of displays with varying peak brightness capabilities (e.g., 600 cd/m², 1000 cd/m², 4000 cd/m²). To ensure consistency, the concept of "headroom interpolation" becomes important. This involves how the signal adapts to displays that may not reach the mastering monitor's full brightness. For example, if a master is created at 1000 cd/m², but a consumer display can only reach 600 cd/m², the HDR content needs to be remapped.

Modern color management systems and delivery formats (like Dolby Vision) handle this by interpolating colors in absolute D65 XYZ space. For example, a 1-stop headroom above reference white can be mapped consistently in XYZ space by color management systems, helping maintain predictable tone mapping across displays.

A common mistake is treating all scenes uniformly, applying a global "HDR look" without considering the individual characteristics of each shot. Pushing highlights globally, for instance, can lead to beautiful results in a bright exterior but can cause blocky or unnatural appearances in a dark, moody interior when viewed on a consumer TV that struggles with low-light detail. Each scene requires careful consideration, and the HDR grade should enhance, not overpower, the original photographic intent. Previewing the OOTF (Opto-Optical Transfer Function) for consumer adaptation is a critical step, allowing you to see how your HDR master will look on displays with different brightness capabilities.

This helps catch potential issues before delivery.

Parallel processing trees are a sophisticated technique used by experienced colorists. One tree might be dedicated to scene-referred creative adjustments (e.g., color warper on mid-tones), while another parallel tree handles display-referred compliance checks and output transforms. This separation allows for maximum creative freedom while ensuring the final output adheres to technical specifications for various delivery platforms.

HDR Delivery Formats: HDR10 vs Dolby Vision Implementation

The final stage of the HDR workflow involves delivering the content in a format suitable for distribution. The two predominant HDR delivery formats are HDR10 and Dolby Vision, each with distinct characteristics and implications for the grading process.

HDR10 is an open standard, making it widely adopted across consumer electronics and streaming platforms. It uses a 10-bit color depth and the BT.2020 color gamut, encoding video using the PQ EOTF. The key characteristic of HDR10 is its use of static metadata. This metadata provides information about the maximum brightness (MaxFALL, Maximum Frame Average Light Level) and maximum light level (MaxCLL, Maximum Content Light Level) of the entire program. This means that a single set of instructions is applied to the entire piece of content, regardless of scene-to-scene variations. While ubiquitous and free to implement, this static nature can be a limitation.

If a film has wildly varying scenes (e.g., a bright outdoor scene followed by a dark night interior), the static metadata might not optimize each scene perfectly, potentially leading to compromises in how the content is rendered on different displays.

Dolby Vision, in contrast, is a proprietary format that offers significant advantages, particularly for premium content. It supports up to 12-bit color depth, the BT.2020 gamut, and the PQ EOTF, but its distinguishing feature is dynamic metadata. This metadata can be generated on a scene-by-scene or even frame-by-frame basis. This allows for highly nuanced tone mapping, where the display's capabilities are dynamically optimized for each moment of the content. For example, an explosion scene might use the display's peak brightness, while a subtle night scene can be carefully adapted to preserve shadow detail without appearing overly bright.

Dolby Vision also specifies support for up to 10,000 cd/m², future-proofing content for displays that may eventually reach those levels.

💡 Pro Tip: Dolby Vision for Dynamic Content

For content with significant variations in brightness and mood between scenes (e.g., action sequences mixed with intimate dialogue), opt for Dolby Vision if possible. Its dynamic metadata allows for frame-accurate tone mapping, ensuring that each scene is optimized for the viewer's display without global compromises.

The implementation of these formats often involves specific software tools. DaVinci Resolve Studio, for example, offers integrated Dolby Vision mastering tools that allow colorists to create the necessary dynamic metadata. This process typically involves setting "targets" for various display brightness levels (e.g., 100 cd/m² for SDR, 600 cd/m², 1000 cd/m² for HDR), and the software helps generate the instructions for how the content should be remapped for each. For HDR10, the process is simpler, primarily involving setting the static metadata values during export.

A common mistake is using HDR10 for highly dynamic content where Dolby Vision's frame-specific tone mapping would provide a superior viewing experience. While HDR10 is a good baseline, it can force compromises when a scene-by-scene adjustment is truly needed. Another pitfall, particularly for productions targeting platforms like Netflix, is attempting to deliver Dolby Vision without a certified facility and workflow. Netflix and other premium platforms have stringent technical specifications and require specific calibration and quality control processes for Dolby Vision content. Dolby itself pre-calibrates certified displays to ensure consistency, which is a significant part of their ecosystem's strength.

Ultimately, the choice between HDR10 and Dolby Vision often comes down to budget, distribution requirements, and the creative ambition for the project. Many productions deliver both, with Dolby Vision serving as the premium offering and HDR10 as a widely compatible fallback. For more context on shooting with HDR delivery in mind, see The Complete Guide to Shooting for HDR and Dolby Vision.

Common Pitfalls, Pro Tips, and Workflow Optimization

Navigating the complexities of HDR requires vigilance and a disciplined approach to workflow. Even experienced filmmakers can fall into common traps if not careful.

One of the most frequent pitfalls is making minor adjustments in HDR without constantly previewing the result on a properly calibrated reference monitor. Because HDR offers such an expanded range, even small tweaks to highlights or shadows can have a highly visible and sometimes unintended impact. A slight push in highlights might introduce clipping, or a subtle lift in shadows could reveal unwanted noise, far more noticeably than in SDR. The solution is constant monitoring and a critical eye on the scopes.

Another significant mistake, as discussed, is attempting to grade HDR on an SDR monitor. An SDR display, typically limited to 100 cd/m², simply cannot represent the full dynamic range of HDR. This leads to a situation where the grader is effectively "guessing" how the highlights and shadows will appear, often resulting in clipped whites or crushed blacks in the final HDR master. Always invest in, or secure access to, a professional HDR reference monitor that meets industry standards for peak brightness, black levels, and color gamut coverage.

💡 Pro Tip: Separate HDR and Contrast Nodes Early

To avoid blocky or unnatural-looking results, especially in challenging scenes, separate your HDR adjustments (e.g., highlight roll-off, shadow detail) from your general contrast adjustments into different nodes, and place them early in your node tree. This allows you to sculpt the dynamic range before applying more aggressive creative contrast, providing greater control and preventing detail loss.

For workflow optimization, a full color pipeline that embraces scene-referred ingest and a display-referred grade is essential. This means starting with footage in its native log or raw format, applying an input transform to a wide gamut, high dynamic range working space (like Rec.2100 PQ). All creative grading should occur in this space. Then, at the very end of the signal chain, an output transform adapts the image for specific delivery targets (e.g., HDR10, Dolby Vision, SDR Rec.709). This structured approach, often facilitated by a well-designed LUT pipeline, ensures consistency and allows for easy creation of multiple deliverables from a single master grade.

Professional colorists often maintain separate node trees for different types of adjustments. For instance, one branch might handle primary color balance and exposure (scene-referred), another for creative "looks" (also mostly scene-referred), and a final branch for display-referred optimizations and compliance checks. This modular approach makes it easier to troubleshoot, make revisions, and adapt the grade for different outputs. Using 12-bit Dolby Vision for future-proofing while also generating an HDR10 fallback is an increasingly common strategy for high-end productions, as it offers the best of both worlds.

Regular calibration of all monitoring equipment, using tools like the X-Rite i1Display Pro, is non-negotiable. Display drift can occur over time, and an uncalibrated monitor undermines the entire grading process. Finally, always test your HDR masters on a variety of consumer devices if possible. While a reference monitor is crucial for mastering, understanding how the content translates to average consumer screens (e.g., different HDR TVs, streaming boxes) can reveal unexpected issues and help refine the grade for broader appeal.

Interface & Handoff Notes

Upstream Inputs

Downstream Outputs

Top 3 Failure Modes for HDR Strategy

1. Grading on Uncalibrated or Non-HDR Monitors: Making creative decisions without an accurate reference display leads to wildly inconsistent results on delivery, with clipped highlights, crushed shadows, or inaccurate colors.

2. Premature Display-Referred Commitments: Applying display LUTs or making creative judgments based on a specific display's limitations too early in the process, thereby losing the flexibility and dynamic range inherent in scene-referred footage.

3. Ignoring Metadata for Delivery: Failing to properly generate or embed static (HDR10) or dynamic (Dolby Vision) metadata, resulting in suboptimal playback on consumer devices and potential rejection by distribution platforms.

Browse This Cluster

- Color Grading Mastery: From Technical Foundations to Creative Excellence (Pillar)

Next Steps

Ready to see how this fits into the bigger picture? Start with the complete guide.

📚 Pillar Guide: Color Grading Mastery: From Technical Foundations to Creative Excellence

---

---

© 2026 BlockReel DAO. All rights reserved. Licensed under CC BY-NC-ND 4.0 • No AI Training.