K-nearest neighbors improves color grading of shadows and highlights
Adaptive Color Grading
Computer Vision and Pattern Recognition
Summary
Adjusting different brightness areas like shadows and highlights helps artists and photographers enhance pictures, but it often relies on simple thresholds that can be tricky to set. The authors created a free color grading tool and used it to label many video frames with brightness thresholds. They tested different prediction methods and found that a k-nearest neighbors method works better than other advanced techniques for applying styles to photos and videos. This shows that focusing on a few key settings can be better than complicated models for creative color editing.
What this means in practice
- •For photo editing software teams: Integrate compact adaptive threshold prediction methods to improve automated color grading of images based on tonescale regions.
- •For video post-production studios: Use the open source color grading tool and dataset to develop better local image manipulation workflows tailored to video frames.
Authors
Trevor D. Canham, Abhijith Punnappurath, Michael S. Brown
Abstract
Independent control of tonescale regions (e.g., shadows, highlights) is essential for painters, photographers and cinematographers to bring 2D images to life. In image manipulation software this is most directly addressed by color grading modules, which use intensity thresholds to segment distinct illumination regions for local manipulation. In this work we develop an open source color grading tool and use it to annotate a large dataset of video frames with tonescale region thresholds. Using these thresholds we conduct modeling experiments with strategies based on both practitioners' conventional wisdom and machine learning. Results show that K-nearest neighbors is an effective prediction strategy, outperforming state-of-the-art end-to-end methods for image enhancement. This outcome demonstrates the benefit of focusing on a compact set of core parameters when modeling creative stylization processes. Our adaptive color grading interface and data are available at https://github.com/SamsungLabs/adaptive-color-grading.