Temporal Image Fusion

TL;DR

This paper presents a novel temporal image fusion technique that enhances dynamic structures and enables long-exposure effects on videos, addressing sensor limitations and expanding creative possibilities.

Contribution

It introduces a new method for temporal image fusion that extends exposure fusion to full-frame videos and long-exposure image generation, handling temporal under-exposure effectively.

Findings

Enables rendering of long-exposure effects on full-frame videos

Addresses temporal under-exposure by enhancing dynamic structures

Produces striking visual effects with content-selective filters

Abstract

This paper introduces temporal image fusion. The proposed technique builds upon previous research in exposure fusion and expands it to deal with the limited Temporal Dynamic Range of existing sensors and camera technologies. In particular, temporal image fusion enables the rendering of long-exposure effects on full frame-rate video, as well as the generation of arbitrarily long exposures from a sequence of images of the same scene taken over time. We explore the problem of temporal under-exposure, and show how it can be addressed by selectively enhancing dynamic structure. Finally, we show that the use of temporal image fusion together with content-selective image filters can produce a range of striking visual effects on a given input sequence.