Learn the Latest About Color

 

Joseph Goldstone has been working on digital representations of color since a summer internship at IBM in 1979. He eventually found himself in the mid-1990s in Los Angeles running the film scanning and recording department at Digital Domain at the time that studio was providing the visual effects for Apollo 13 and Titanic (the two that have most stood the test of time). He then went north to San Francisco and worked in the Film Group within Industrial Light & Magic, experimenting with early film color management; this work was presented at CIC11 in 2003.

Goldstone was part of the team that developed the ACES color management substrate, for which the group was awarded an Engineering Emmy in 2012. From 2011 through 2024, he contributed to the Image Science group at ARRI Cine Technik in Munich, while being based in Los Angeles and thus proximate to the major US motion picture studios. During that time, he contributed to SMPTE standards relevant to HDR mastering and final display.

He is a SMPTE Fellow, a recipient of the SMPTE Technicolor Herbert T. Kalmus Medal, an associate member of the American Society of Cinematographers, and a member of IATSE Local 600. Goldstone wrote the color management chapter of the Visual Effects Society's Visual Effects Handbook; and he serves on investigatory committees for the Motion Picture Academy’s Science and Technology Awards.

After several years spent contributing to the development of the output format used by ARRI’s flagship camera, he is very pleased to return to the field of color management for visual effects, to learn what the cool kids have come up with while he was otherwise engaged.

Panoramas are formed by stitching together two or more images of a scene viewed from different positions. Part of the solution to this stitching problem is ‘solving for the homography’: where the detail in one image is geometrically warped so it appears in the coordinate frame of another. In this paper, we view the spectral loci for a given camera and the human visual system (i.e. their respective chromaticity diagrams) as two pictures of the same ‘scene’ and warp one to the other by finding the best homography. When this geometric distortion renders the two loci to be identical then there exists a unique colour filter (that falls gracefully from the derivation without further calculation) which makes the camera colorimetric (the camera+filter measures RGBs that are exactly linearly related to XYZs). When the best homography is not exact the filter derived by this method still makes cameras approximately colorimetric. Experiments validate our method.

 

Yannick Hold-Geoffroy is a senior engineer and research scientist working at Adobe since 2018. He focuses on lighting estimation and modeling, inverse rendering, image analysis, and 3D reconstruction. He earned his PhD from Laval University, Québec City, under the supervision of Jean-François Lalonde. Hold-Geoffroy notably contributed to Adobe's image generator Firefly, the Match Image feature in Adobe Stager, and the infrastructure behind Neural Filters in Photoshop, among other things. His passion lies in creating new technologies that simplify technical work, allowing multimedia artists to concentrate on visual storytelling.

The colorimetric calibration of cameras are critical in imaging systems, with the sources used in light booths being widely used in practice. These sources, however, may not good presentations of the sources in real life, which possibly results in poor colors. In this study, we adopted a genetic algorithm and a large dataset of real light sources to identify an optimal set of sources that can better represent the sources in real life. The experiment results suggested that the identified set of sources can result in better color performance. Moreover, the selection of the sources was much less complicated in comparison to manual selections, which can be considered and implemented in practice.

Preserving perceptual quality of the tone mapped images is one of the major challenges in tone mapping. Most traditional tone mapping operators (TMOs) compress the luminance of high dynamic range (HDR) images without taking account of image color information, resulting into less natural or preferable colors. Current color management algorithms require either manual fine-tuning or introduce lightness and hue shifts. An adaptive color correction model is proposed to address color distortions in tone mapping. It is based on the CIECAM16 to compute perceptual correlates, i.e., Lightness, Chroma and Hue. Regardless of the tone mapping technique, the proposed model recovers natural colors of tone mapped images for spatially invariant and variant operators, making it an effective post-processing technique for color reproduction. Unlike other models, it requires no gamut mapping correction, reproducing more accurate hue, chroma, and lightness. The algorithm was evaluated using objective and subjective methods, revealing that it produced significantly better color reproduction for tone mapped images in terms of naturalness of the colors.

Colors are characterized by their appearance attributes such as brightness, colorfulness and hue, or lightness, chroma and hue when considered relative to an adapted white. Consequently, they can be represented by coordinates in three-dimensional spaces like those of CIELAB or CIECAM02 appearance predictors. Therefore, the color of a stimulus is customarily associated with a point in 3D. While this is appropriate when dealing with theoretical quantities, in practice the color space coordinates of a stimulus are subject to variations both in the stimulus itself and in its measurements. This tends to lead to multiple measurements being the basis of identifying the color of a stimulus by averaging them and in some cases excluding outliers. This, however, obscures the fundamental variability of color data. In this paper an alternative, probabilistic approach to dealing with color data will be introduced and applied to the computation of color difference, color gamuts and gamut inclusion, yielding distributions of answers instead of only single values.

The Monk Skin Tone (MST) scale, comprising ten digitally defined colors, represents a diverse range of skin tones and is utilised by Google to promote social equity. The MST scale also has the potential to address health disparities in the medical field. However, these colors must be printed as physical charts for health practitioners and researchers to conduct visual comparisons. This study examines the colorimetric characteristics of the MST scale and establishes two per-patch acceptance criteria to determine the acceptability of a printed MST chart using a four-tier grading method. A software tool was developed to assist end users in printing accurate MST charts. The tool was tested with four printer/paper combinations, including inkjet, dye sublimation, and laser printers. Results show that a wide color gamut covering the brightest and darkest MST levels is crucial for producing an accurate MST chart, which can be achieved with a consumer-grade inkjet printer and glossy paper.

Human perception of color varies between individuals, raising the question of how well the standard color matching functions (CMFs) represent individual observers in image reproduction. The goal of this research is to explore the relationship between CMFs and both fidelity and preference in the image reproduction pipeline. Three experiments were conducted: an experiment to estimate approximately individualized CMFs, an image fidelity experiment, and an image preference experiment. The results show that the CMFs influence the accuracy of image reproduction, however, preferences are affected by factors in addition to CMFs. The findings offer insights into the limitations and potential implications of relying on using standard CMFs in image reproduction technologies.

Biophysical skin appearance modeling has previously focused on spectral absorption and scattering due to chromophores in various skin layers. In this work, we extend recent practical skin appearance measurement methods employing RGB illumination to provide a novel estimate of skin fluorescence, as well as direct measurements of two parameters related to blood distribution in skin – blood volume fraction, and blood oxygenation. The proposed method involves the acquisition of RGB facial skin reflectance responses under RGB illumination produced by regular desktop LCD screens. Unlike previous works that have employed hyperspectral imaging for this purpose, we demonstrate successful isolation of elastin-related fluorescence, as well as blood distributions in capillaries and veins using our practical RGB imaging procedure.

In college at ODTU Ankara, Qasim Zaidi started by studying economics but shifted to mathematics. In grad school at the University of Chicago he studied behavioral decision making and visual neuroscience and did a post-doc in color perception at Bell Labs. In his lab at Columbia University, he used newly developed graphics systems to study higher level color and motion mechanisms, with applications to glaucoma, diabetes, and other retinal abnormalities. Zaidi moved to SUNY to build the Graduate Center for Vision Research. His lab does experimental and computational work on perception of rigid and non-rigid 3-D shapes and materials, contextual effects on color perception, and form distortions seen by amblyopes. In his primate neurophysiology collaborations, he uses connectomics, patch clamping and 2-photon imaging to understand the wiring and dynamics of different classes of retinal ganglion cells, fMRI guided micro-electrode recordings to understand color decoding by inferotemporal neurons, multi-electrode recordings and mathematical models to understand the structure of primary visual cortex (V1), and retrograde injections and optical imaging to understand how anatomically identified large-scale inputs from V1 to individual V2 cells are combined to give emergent properties. Zaidi's research is made possible by multiple continuing grants from NIH.

This article describes two color matching experiments using 6 displays and a visual trichromator based on spectral tunable LED technology respectively. Ten observers performed the experiment. The results were used to reveal observer metamerism between displays, to derive each individual’s color matching functions (CMFs), and to test their performance against the available CMFs. Comparing different CMFs to estimate observer metamerism of displays, individual CMF (expressed by mean ∆E₀₀ of 2.7) clearly outperformed the others. Amongst the CIE CMFs, CIE 2006 10° performed the best (5.1), followed by the other color matching functions, CIE 1931 2° gave the worst performance (11.1). The results indicates a large error could occur by adopting the current standard (CIE 1931 2°) to specify colors and this can be much improved using the individual CMFs. A simulation analysis of display primaries was also conducted to show peak wavelength to have a greater impact on observer metamerism than peak bandwidth.

As VR technology has advanced, its use in performance-critical fields such as medical training and vision research has grown, driving a need for increasingly realistic VR environments. In previous work, we evaluated lightness constancy in a task where viewers matched the reflectance of surfaces at different 3D orientations, and we found substantially poorer lightness constancy in VR than in a physical apparatus. Poor constancy in VR may have been due to simplified rendering of scenes in that study, e.g., largely achromatic Lambertian surfaces. Motivated by these findings, here we evaluated lightness constancy in more realistic VR scenes, rendered with a broad array of materials, colors, textures, and specular highlights, as well as more realistic shadows. We tested two conditions: a Full-Context condition, where these lighting and material cues were available, and a Reduced-Context condition, where they were not. Participants had significantly better lightness constancy in the Full-Context condition than in the Reduced-Context condition, indicating that they exploited these additional cues. However, lightness constancy was still quite poor in absolute terms, despite the availability of rich lighting and material cues. The reasons for this failure of constancy are unclear from previous literature, and this finding suggests a promising research problem with both fundamental interest and practical applications.

3-dimensional translucent objects exhibit large spatial variations of color, and it remains unclear what colors humans choose cognitively to represent monomaterial 3D translucent objects. Previous works asked human observers to match translucent object with a most representative uniform color patch. Matched colors were compared with the average color of the object, which turned out to be a poor predictor of the match. In this work, we conducted a more thorough analysis of the color matching experiments and investigated whether matched color is systematically overestimated or underestimated relative to the average color in any particular axis in the CIELAB space. Afterward, we conducted K-means clustering of the pixel colors to extract a palette of dominant colors of the translucent object and compare the palette with observers’ match. And finally, we went through a psychophysical experiment to quantify perceived color differences between pairs of 3D translucent objects. We explore whether perceptual color differences can be predicted with traditional color difference formulae using pixel-wise averages or dominant colors from the K-means clustering.