Tone Mapping High Dynamic Range Videos using Wavelets

Qi Shan, Mark Meyer, Tony DeRose, John Anderson

Abstract:

We propose a novel 3D wavelet based tone mapping framework for high dynamic range videos. Still image tone mapping methods can be applied to videos in a frame by frame fashion, but they often exhibit haloing artifacts and do not guarantee temporal coherence, resulting in flickering. Directly extending wavelet analysis/synthesis to time solves the flickering problem but leads to ghosting. Our method, based on a new variant of edge avoiding wavelets in time, addresses both flickering and ghosting. We also reduce haloing artifacts through the use of a new gain control method. Our method produces significantly higher quality results compared to existing state-of-the-art techniques when applied to both live action and computer graphics video.

Available as Pixar Technical Memo #12-01