Color Science for Developers: Understanding LUTs

Look Up Tables (LUTs) are fundamental to modern color grading, yet their inner workings remain mysterious to many developers. Let’s demystify them by building a color grading pipeline from scratch.

What is a LUT?

A LUT is essentially a function that maps input colors to output colors. Think of it as a giant lookup table where you can find any RGB input value and see what output value it should become.

The Mathematics

A 3D LUT creates a cubic lattice in color space. Each point in the cube represents a specific RGB combination, and the values stored at that point define the transformation.

import numpy as np

def apply_lut(image, lut, size=33):
    """Apply a 3D LUT to an image."""
    # Normalize to 0-1 range
    normalized = image / 255.0
    
    # Scale to LUT indices
    indices = normalized * (size - 1)
    
    # Trilinear interpolation for smooth results
    return trilinear_interpolate(lut, indices)

Building Cinegrade

This understanding led me to build Cinegrade, an open-source tool that generates custom LUTs based on film stock characteristics. The key insight was treating historical film data as training data for the color transformations.