enhancer

A C++/GLSL library for enhancing photographs (adjusting brightness, contrast, etc.).

This repository contains the following three features:

• GLSL shaders: Enhancement functionality as shaders for real-time enhancement applications.
• C++ functions: Enhancement functionality as C++ functions for display-less environments.
• Qt Widget: Utility Qt-based widget for easing the use of the GLSL shaders.

Supported Parameters

Default Set

• Brightness:
  • Apply a gamma curve in the RGB space
• Contrast:
  • Apply an S curve in the (gamma-corrected) RGB space
• Saturation:
  • Scale the saturation
• Temperature:
  • Add an offset toward the -U+V direction in the Y'UV space
• Tint:
  • Add an offset toward the +U+V direction in the Y'UV space

Optional

• Lift/Gamma/Gain (9D)
  • Reproduce common effects used in many color grading software packages
• Color Balance (3D)
  • Adjust the RGB values while preserving the lightness

Examples

Brightness

Contrast

Saturation

Temperature

Tint

Required Runtime Environments

• GLSL 3.3
• OpenGL 3.2 Core Profile (for Qt features only)

Dependencies

C++ Features

• Eigen (for macOS users, brew install eigen; for Ubuntu18.04 users, apt-get install libeigen3-dev)

C++ Qt Features

• Qt6 (for macOS users, brew install qt) or Qt5 (for Ubuntu 18.04 users, apt-get install qt5-default)

C++ API

Eigen::Vector3d enhance(const Eigen::Vector3d& input_rgb,
                        const Eigen::VectorXd& parameters);

where input_rgb is a 3-dimensional vector ([0, 1]^3), and parameters is a 5-dimensional vector ([0, 1]^5).

Projects using enhancer

• Sequential Gallery [SIGGRAPH 2020] https://github.com/yuki-koyama/sequential-gallery
• Sequential Line Search [SIGGRAPH 2017] https://github.com/yuki-koyama/sequential-line-search
• SelPh [CHI 2016] https://github.com/yuki-koyama/selph

Note: About Older Version (v1)

This library used to have a different procedure to enhance photographs, which was used for some previous publications [Koyama+UIST14; Koyama+CHI16; Koyama+SIGGRAPH17]. The original procedure has color_balance parameters (3D) instead of temperature and tint (2D). Thus, the total number of parameters was 6, not 5.

The color_balance parameters are for manipulating the chrominance of the input color, which is inherently 2D, and so assigning 3 parameters for this purpose is considered redundant. For example, color_balance = [0.5, 0.5, 0.5] and color_balance = [0.2, 0.2, 0.2] result in the same color even though the parameter sets are different.

Thus, the enhancement procedure was updated to use temperature and tint instead of color_balance. These two parameters are popularly used in Adobe Photoshop and Lightroom and so experts are likely to familiar with the concept. These two parameters also appeared in [Kapoor+IJCV14].