[其它] Capturing Shape and Reflectance of Food

[复制链接]

彬彬

797 主题	1 听众	1万积分

资深设计师

Rank: 7 Rank: 7 Rank: 7

纳金币: 5568
精华: 0

电梯直达

楼主

发表于 2012-1-5 08:39:59 |只看该作者 |倒序浏览

Abstract

Photo-realistic 3D content is crucial for creating convincing digital

images. For certain classes of everyday objects, such as food, the

human perception is sensitive to even small inconsistencies, making

the creation of such content difficult and time-consuming, even for

experts. In this sketch, we will explore the use of an automated

pipeline for capturing 3D shape and Bidirectional Texture Function

of food. The acquired data is used to render photo-realistic images

of purely virtual objects under arbitrary lighting and with full global

illumination.

CR Categories: I.3.3 [Computer Graphics]: Picture/Image

Generation—Digitizing and scanning; I.3.7 [Computer Graphics]:

Three-Dimensional Graphics and Realism—Color, shading, shadowing,

and texture; I.4.1 [Image Processing and Computer Vision]:

Digitization and Image Capture—Reflectance;

Keywords: BTF, food visualization, image-based rendering

Links: DL PDF WEB VIDEO

1 Introduction

The major bottleneck for generating convincing photo-realistic

images is the creation of digital 3D content including 3D models

and reflectance properties.

For many classes of objects this creation process is far from being

trivial and usually requires a tremendous amount of manual work

by expert 3D artists.

The 3D modeling of fresh food is certainly an especially challenging

example. Humans have developed a high degree of sensitivity

to this subject matter, and, hence, even slightest errors may yield

visually unconvincing results. One approach for obtaining realistic

3D content is modeling by example, where existing real-world

objects are captured.

In many cases, it is desirable that the generated content can be used

in arbitrary synthetic scenes with possibly multiple objects, novel

viewpoints and novel lighting. This is an inherently challenging

task, since both a full geometric representation as well as spatially

and angularly varying reflectance properties need to be acquired.

With application to food modeling in mind, capture times have to

be considered as well, limiting the range of applicable approaches.

In this sketch, we explore the use of a parallelized system for automated

capturing of geometry and Bidirectional Texture Function

(BTF) of food. Once a model has been acquired it can be used

to create photo-realistic images showing accurate shadowing and

global illumination for arbitrary lighting conditions.

Our results indicate that the proposed pipeline is able to handle even

optically complex cases (Fig. 1) that normally require extremely

costly manual modeling.

2 Our Pipeline

Our goal is to capture a fully realistic 3D-representation of fresh

food which imposes serious restrictions on the acquisition pipeline:

As most fresh food degrades quickly, the acquisition process needs

to be fast enough to avoid changes in appearance during measurement.

In addition, food tends to deform over time, which means

that both shape and reflectance have to be captured within a short

time frame, ideally without moving the object.

For meeting the above requirements, we employ a method originally

developed in the context of cultural heritage [Schwartz et al.

2011], which performs an integrated capture of geometry and BTF

and that benefits from extensive parallelization. Unlike serial setups

(a recent example is [Holroyd et al. 2010]) a dense acquisition of reflectance

is possible, capturing complex detail, such as mesostructure,

interreflections, local sub-surface scattering and specularity.

The capturing system. Our setup (Fig. 2) uses a hemispherical

gantry, with 151 consumer-cameras (Canon PowerShot G9) evenly

covering the hemisphere around the food-sample. In addition, 8

projectors are mounted on the gantry.

To reduce degradation effects, the room containing the setup was

actively cooled to about 16C. Furthermore, we acquired appearance

before geometry, as accuracy of reflectance has a more significant

impact on human perception.