Nonlinear color triads for approximation, learning and direct manipulation of color distributions

Abstract

We present nonlinear color triads, an extension of color gradients able to approximate a variety of natural color distributions that have no standard interactive representation. We derive a method to fit this compact parametric representation to existing images and show its power for tasks such as image editing and compression. Our color triad formulation can also be included in standard deep learning architectures, facilitating further research.

Piet: Facilitating Color Authoring for Motion Graphics Video

“…However, the palette exploration and its assignment to the intended design remain separate. Bridging this gap, tools like Color Triads [29] and Histomages [7] introduced a triangular-shaped palette and color histogram, respectively, to allow users to directly recolor images by manipulating the palette. Specifically tailored for finegrained color blending, Playful Palette [30] employs a free-form unstructured palette for artwork colorization.…”

“…-E2, a loop that includes communications with clients and iterative refinements. [29] revealed that color design is an iterative process in static graphic designs. Our findings dive into the color design process in MG videos, particularly, how designers explore the color choices in an early stage and how they make necessary adjustments later.…”

“…Despite the extensive studies in HCI and media have investigated approaches to animation authoring [14,17], exploration of visual storytelling [19,23], and the broader difficulties faced in motion graphics creation [15], the specific process and challenges of color design in MG videos remain largely unexplored. Despite the wealth of research addressing the challenges and opportunities in static graphics colorization [7,16,29,30], the knowledge gap in MG video color design and authoring is evident. Currently, there is a limited understanding of the color authoring process in MG videos and the specific challenges designers face during this process.…”

PosterChild: Blend‐Aware Artistic Posterization

“…Rang et al [RPCB17] developed a modified 2D clustering algorithm based on the ab channels of LAB‐space for extracting themes and recoloring images. Shugrina et al [SKFS20] presented a minimal nonlinear primitive to model images with 2‐manifold color distributions. These approaches create palettes with custom (i.e., non‐linear) blending or editing approaches.…”

HistoGAN: Controlling Colors of GAN-Generated and Real Images via Color Histograms

“…These color transfer methods aim to manipulate the colors in an input image to match those of a target image, such that the images share a similar "look and feel". In the color transfer literature, there are various forms of color histograms used to represent the color distribution of an image, such as a direct 3D histogram [20,51,62], color palettes [7,12,64] or color triads [58]. Despite the effectiveness of color histograms for color transfer, recent deep learning methods almost exclusively rely on image-based examples to control colors.…”