Hair Color Digitization Through Imaging And Deep Inverse Graphics

The study introduced a novel method for hair color digitization using a combination of a controlled imaging device, a path-tracing renderer, and an inverse graphics model based on self-supervised machine learning. This approach allowed for capturing the color appearance of physical hair samples and rendering synthetic images with similar appearances, simulating various hairstyles and lighting environments. Unlike conventional methods, this did not require differentiable rendering for training. The method demonstrated accurate capture and rendering of hair color on both real and synthetic images, offering potential applications in augmented/virtual reality and hair dye development.