[其它] Estimating Dual-scale Properties of Glossy Surfaces from Step-edge Lighting

[复制链接]

晃晃

1023 主题	3 听众	359 积分

设计实习生

Rank: 2

纳金币: 335582
精华: 0

电梯直达

楼主

发表于 2011-12-29 08:44:44 |只看该作者 |倒序浏览

Estimating Dual-scale Properties of Glossy Surfaces from Step-edge Lighting

Chun-Po Wang Noah Snavelyy Steve Marschner

Cornell University Cornell University Cornell University

Abstract

This paper introduces a rapid appearance capture method suited for

a variety of common indoor surfaces, in which a single photograph

of the reﬂection of a step edge is used to estimate both a BRDF and

a statistical model for visible surface geometry, or mesostructure. It

is applicable to surfaces with statistically stationary variation in sur-

face height, even when these variations are large enough to produce

visible texture in the image. Results are shown from a prototype

system using a separate camera and LCD, demonstrating good vi-

sual matches for a range of man-made indoor materials.

Keywords: appearance capture, reﬂectance, rendering

1 Introduction

Acquiring the appearance of real-world surfaces is important when

real scenes need to be modeled faithfully by computer graphics.

However, even the near-homogeneous manufactured surfaces—

paints, metals, plastics—that are ubiquitous in man-made environ-

ments have evaded easy measurement. With the assumption of ho-

mogeneity, surfaces can be described by a single reﬂectance func-

tion that can be estimated from images [Yu et al. 1999] or rapidly

acquired by handheld devices [Dong et al. 2010; X-Rite 2011], but

the resulting surfaces always look too featureless, because surfaces

with no visible texture are rare. Good, detailed appearance can be

achieved by measuring high-resolution parameter maps for particu-

lar samples [Gardner et al. 2003; Ghosh et al. 2009; Ghosh et al.

2010], but these methods are considerably less convenient—too

much work to add subtle texture to a basically homogeneous sur-

face.

In this paper we explore a middle path, suited for many common

indoor surfaces, in which a rapid single-image measurement results

in a statistical description of both visible and microscopic surface

roughness that is sufﬁcient to yield renderings of a surface that qual-

itatively match its appearance. Figure 2 shows some examples of

surfaces that appear in a typical indoor scene. Note that these sur-

faces can be described not only in terms of reﬂectance—the diffuse-

ness of the wall, the gloss evident in the reﬂection off the cabinet—

but also by visible “bumps” with characteristic frequency content.

Both of these phenomena are important to the appearance of sur-

faces (particularly in high-resolution imagery), and our approach

explicitly handles both. However, our goal is not to measure ex-

act surface properties—full BRDFs, accurate normal maps—but in-

stead to capture enough information about the statistics of a surface

to achieve a qualitative appearance match. To this end, we propose

a dual-level model of surface appearance, with one level model-

ing microscopic surface geometry (described by the BRDF), and

another modeling visible surface bumps (the mesostructure often

represented in a normal or bump map). To estimate the parameters

of our model, we propose an appearance capture system that uses a

single image of the reﬂection of step-edge illumination from a pla-

nar sample of a surface to estimate both the BRDF and the statistics

of meso-scale geometry (Figure 1). While we present a prototype

capture system using a separate camera and LCD, our system could

potentially be implemented on a consumer device with a display

and a front-facing camera, such as an Apple iPad, and eventually

could use natural illumination (which often includes step edges).

Our appearance model and capture system are designed as a whole

with the goal of enabling very simple, robust acquisition, while still

handling many interesting real-world surfaces. We begin by brieﬂy

describing these two components.

全文请下载附件：