Challenges in Modern Content Creation

The Colorfront Engine™ (CFE) is a color processing system based on human visual perception, with applications ranging from handling camera original images to remapping finished masters.

Industry Practices

Understanding the Continuum of Dynamic Ranges
Read More

When considering current display technologies, we see a great variation in dynamic range, color gamut, and the recommended (or realistic) display environments in which content is viewed. It is a mistake to think about these displays as either HDR or SDR in a binary way. In reality, there is a continuum of different dynamic ranges between the two. There is no magic number defining where HDR begins and SDR ends.

For example, a 48-nit cinema version has lower peak brightness than a 100-nit broadcast deliverable. Does that mean 48-nit cinema is a sub-SDR if we call 100-nit SDR? Regardless of labeling them both SDR, they are simply displays with different capabilities, rendered in different viewing environments. Cinema venues continue to expand their offerings of enhanced and high dynamic range technologies, from laser projectors to emissive LED walls. By achieving very deep blacks levels in dark cinema environments, consumers are treated to premium experiences with higher contrast ratios.

The Need for a Managed Color Workflow
‍
Read More

That said, how do we master each of these different versions? Do we have a colorist producing every single one of these deliveries? To avoid the need to grade for each display, a managed color workflow is required that translates between different nit levels and color spaces while maintaining a perceptual match. In other words, when viewers look at the content, they need to see the same thing.

In an ideal situation, a reference master display should be a superset of all of the deliverables. For instance, if our maximum deliverable reaches 1000 nits, we should be able to visualize how all other possible deliverables will appear. However, this is often not possible. Instead, we may be limited to the existing viewing technology and unable to exceed its capabilities. To address this, we need systems that can translate from our particular arrangement to displays with different brightness levels.

Preserving Creative Intent Across Various Deliverables
Read More

Using a reference master with very high peak brightness presents its own challenges, as mapping to a lower dynamic range display with a possibly darker ambient surround is not straightforward.

Without the right tools, it is easy to diverge from the original creative intent, resulting in an unfortunately altered look across various deliverables. Below, we discuss some typical use cases and current approaches that attempt to overcome these challenges.

Dynamic Range

Dynamic Range of Of Reference Master Larger Than Deliverable.

A typical use case scenario arises when the reference master’s dynamic range is high, graded on a brighter display, while a deliverable with a lower dynamic range is needed. For example, we may have a master graded on a bright 1000-nit HDR display that needs to be delivered to a cinema with a 48-nit display.
Scaling Down Images and Brightness Loss
There are various ways to scale down the image and reduce brightness, but they all come with some loss of the full-range image. Clipping highlights produces poor image quality at the clipping points, while soft clipping also results in the loss of important tonal and color relationships.

A more complex tone curve could be a better solution; however, color hue, saturation, and colorfulness may not match accurately with this approach either. Although these techniques could work reasonably well with black-and-white images, they are not ideal in most present-day cases where we work with color images.
Dolby Vision for HDR to SDR Conversion
Dolby Vision is one widely adopted technology for mapping HDR content to SDR, offering a comprehensive workflow to handle high-to-low dynamic range conversions. However, the Dolby Visionworkflow requires both statistical pre-processing and scene change analysis, which adds to the overall complexity.

This pre-analysis can make the process time-consuming, as it involves ensuring that the intended visual quality is preserved across all scenes. While this approach can deliver impressive results, it often requires a colorist to apply creative trims, which, combined with extensive pre-processing, may hinder the efficient use of creative time.
Grading Down Higher-Range Images
A technique often used is to grade down a higher-range image through color correction. A skilled colorist may then have to spend significant time scaling down the image to fit it into a lower dynamic range format for each scene, often using shapes and secondaries to achieve a visually pleasing result. This process consumes valuable creative time, preventing the colorist from focusing on enhancing the image and applying a unique creative look to the piece.

Consistency must also be maintained across scenes and deliverables, a process that takes a significant amount of time. However, with the proper transform, you can preserve perceived colors and tonal relationships, maintaining skin tones and grayscale, ultimately resulting in a match across various deliverables.
Tab 1
Tab 2
Tab 3

SDR To HDR

Dynamic Range of Deliverable Larger Than Reference Master

This opposite scenario involves mastering a deliverable at a lower nit level than the intended dis-play. For example, starting with a 100-nit reference master and needing to adapt it to HDR. While manually increasing brightness is an option, this approach often results in an overly bright SDR image that lacks the intended aesthetic quality.
A common use case is remapping 48-nit cinema masters for display on newer emissive cinema walls.
Read More

A common use case is remapping 48-nit cinema masters for display on newer emissive cinema walls. Given the time-intensive process of remastering movies with current industry practices, most films shown on these new screens are unfortunately unable to fully utilize the enhanced dynamic range capabilities of these displays.

Manually grading the image to achieve a balanced, visually pleasing brightness can be tedious and labor-intensive.
Read More

Manually grading the image to achieve a balanced, visually pleasing brightness can be tedious and labor-intensive. Another option is to enhance only the highlights; however, this can disrupt color and grayscale relationships, leading to a loss of creative intent. Ideally, we want to maintain these color relationships and preserve the essence of the original image. By processing the content in a perceptual environment, these characteristics are naturally aligned, resulting in a cohesive and well-balanced image.

There are various transforms available to the industry, often fixed 3D Lookup tables.
Read More

There are various transforms available to the industry, often fixed 3D Lookup tables. Many are based on the ITU-R BT.2390 specification which merely places the SDR content inside the HDR container at 203 nits, twice the brightness. This results in just a brighter SDR image. On the other hand, the Colorfront Engine’s Perceptual Processor unbuilds the SDR image, places it into the internal perceptual processing space where it can be reconstructed into a beautiful dynamic HDR image as though it had been originally mastered in HDR, all the while preserving the creative intent of the original SDR image.

SoftPreviewing

Soft proofing, or soft previewing, is essential for verifying color accuracy. We need to preview the content on a high dynamic range display, as well as on all deliverables with varying nit levels.
This capability allows us to soft proof across different dynamic ranges and color constraints, such as 100 nits for Rec.709 tele- vision, or 300 nits for a cinema wall, directly on the reference master display.

It is important that monitor settings remain unchanged: the system should allow us to toggle between deliverables with a single button to ensure consistent grading quality.

Output Display Transform

Once the master look is finalized and rendered to the desired nit level, the output is prepared for transmission systems, or “handshake spaces.” For HDR, the primary transmission systems are HLG and HDR10. When correctly implemented, HLG and HDR10 should reproduce identical colors, ensuring a consistent appearance. For SDR, BT.709 is used for television and XYZDCI is used for cinema. Transmission systems and handshake spaces are designed not to alter the intended look when properly implemented.

Color Gamut Remapping

In cases where the master content has a larger color palette than the transmission system or intended display, colors from the larger color space must be constrained into the smaller one. Simply clipping these out of gamut colors can produce artifacts such as drastic hue shifts, blocking, and other distortions. The goal of proper out-of-gamut remapping is to find the most perceptually accurate substitute colors that fit into the delivery color space specification.

Creative Trimming

Colorfront Engine is equipped with specialized controls to compensate for changes in brightness and viewing conditions while maintaining a perceptual match.
Most industry tools are not designed to preserve color perception accurately and therefore require advanced skills and significant time to achieve satisfactory results for different display nit levels and viewing conditions.
Dynamic range control:
To manage dynamic compression or  expansion within the target nit range.
Ambient surround compensation:
To adjust for different viewing environments, from dark to bright and vice versa.
Colorfulness adjustment:
To regulate the perceived vibrancy of remapped colors.

Limitations

Perceptual Color Transforms.
Current output transforms (ODTs) in the industry are not designed to ensure a perceptual match across different deliveries.
Each transform introduces a unique look, often requiring a separate grade for each delivery to maintain visual quality. This makes preserving the original creative intent across all deliveries challenging and time consuming.
For example, an image may initially be rendered to 100nit Rec.709 with one of the standard ODTs, but later needs to be converted to cinema 48nit DCI P3.

Simply changing the ODT for 48nit DCI P3 may produce a satisfactory result, but without perceptual matching, there will be noticeable differences in color and saturation, which the colorist would then need to correct.

However, using a perceptual transform to create a darker version of the image ensures more accurate color and skin tone reproduction.