Event-Based Shape and Reflectance Acquisition Using the Refresh Line of a Single Off-the-Shelf Display
Year: 2026
Type: article
Abstract: Capturing the shape and spatially varying reflectance of objects is fundamental to many graphics and vision applications, yet achieving fast acquisition remains challenging. Existing display-based appearance capture methods rely on a sequence of multiple illumination patterns and measurements to disentangle illumination, geometry, material response, and sensor characteristics. However, pattern switching and multi-frame capture are time-consuming, limiting acquisition speed and applicability to dynamic scenarios. In this paper, we observe that the native row-wise refresh lines of commodity displays already provide a fast spatiotemporal illumination signal within a single frame. Combining this with a high-temporal-resolution event camera, we capture object appearance changes within one frame during the display's progressive scanning. Based on this signal, we propose a method to simultaneously reconstruct shape and reflectance from the event stream induced by refresh lines. We build a prototype system that pairs a commodity OLED display with an event camera to capture refresh-line responses from near-planar, highly metallic objects. Using this setup, we develop a reconstruction pipeline that uses event measurements to recover surface shape and spatially varying reflectance. Through 25 fps real-time capture and offline reconstruction on real near-planar metallic objects, we demonstrate that native display refresh can be repurposed from an imaging artifact into a practical high-speed cue for appearance acquisition.
Source: ACM Transactions on Graphics
DOI: 10.1145/3842542
Authors: Zhuofeng Wang, Bohan Yu, Jiayi Zhou, Boxin Shi
Institution: Peking University
Language: English
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7225620837