A VIRTUAL ENVIRONMENT FOR ARCHIVING

MICRO-PRESENCE WITH IMAGE-BASED MODEL ACQUISITION

Masahiko Morita

, Tastuya Saito

and Kenji Kohiyama

Graduate School of Media and Governance, Keio University, 5322 Endoh, Fujisawa-shi, Kanagawa, Japan

Tokyo National University of Fine-Arts and Music, 12-8 Ueno Kouen, Taito-ku, Tokyo, Japan

Keywords:

Computer graphics, photography, transparent, transparency, diffusivity.

Abstract:

Micro Archiving is a new method of photography that helps anybody digitally archive minute structures that

existing capturing systems cannot deal with. For research use, it is an effective way to make high-ﬁdelity

virtual specimens of small objects that allowscholarstosharethepreciousacademicresourcesover the Internet

in a digital format. For commercial use, the virtual models captured by the technology can be applied to

computer graphics, games, digital cinemas and other digital entertainment media. Technically, our Micro

Archiving technology enables easy and fast construction of high-quality and high-resolution textured virtual

3D models of small objects. Currently, application contents of Virtual Reality, Augmented Reality and other

interactive media are dependent on the use of elaborate 3D models, which require expensive manual task.

Micro Archiving technology provides automatic capturing process that can comprehensively handle a high-

resolution all-focused imaging, 3-dimensional modeling and transparency capturing simultaneously. Micro

Archiving is best suited for minute and complex objects such as machine parts, insects, plants and small

archaeological materials.

1 INTRODUCTION

Invention of optical devices such as telescopes and

microscopes widened the perceivable spectrum of the

reality of the natural world by allowing us to see what

is invisible to our naked eyes.

Macro

Micro

Hidden

Visible Reality

Micro-Presence

Reality

Hidden

Reality

Figure 1: Micro Presence.

As well as revealing hidden aspect of the world,

archiving it has been a challenge of importance in

research ﬁelds where replicable duplication of spec-

imens and samples plays a crucial role. For example,

in biological studies on a small animals such as in-

sects and microscopic organisms, a photography and

hand-drawn sketches help researchers share scientiﬁc

knowledge on invariable specimens. This applies to

studies on artiﬁcial objects such as archaeological ma-

terials from the ancient time. In ancient sculptures

and fossils, their ﬁne surface structures are historical

and cultural proofs that tell about how the objects are

produced and how they are used. Archiving and in-

vestigating these proofs is an important element of a

research task.

Here, the concept of Micro-Presence is introduced

to mean a kind of reality that is too small to be directly

experienced by our sensorial organs. A microscope is

a good example of a device to unveil Micro-Presence

by providing us with a way to observe microstructures

of both natural and artiﬁcial objects.

Archiving Micro-Presence such as insects, plants,

mineral substances and small-sized archaeological

ﬁndings for research use is often useful in that it al-

lows researchers and archaeologists to observe elabo-

rate virtual samples to inspect physical features with-

out having the actual samples. Use of virtual sam-

ples became a major technical topic as multimedia

database technology has highly advanced in the last

decade and the access to it has became extremely low-

cost. This trend will be accelerated in the future as

145

Morita M., Saito T. and Kohiyama K. (2007).

A VIRTUAL ENVIRONMENT FOR ARCHIVING MICRO-PRESENCE WITH IMAGE-BASED MODEL ACQUISITION.

In Proceedings of the Second International Conference on Computer Graphics Theory and Applications - GM/R, pages 145-152

DOI: 10.5220/0002081101450152

 SciTePress

the bandwidth of the Internet increases and interface

technology becomes easily accessible. Our project,

Micro Archiving, aims at developing a technology to

archive Micro-Presence in highquality. For research

use, it is an effective way to make highﬁdelity vir-

tual specimens that allow scholars to share academic

resources over the Internet. For commercial use, the

virtual models captured by our technology can be ap-

plied to computer graphics, games, digital cinemas

and other entertainment media. The technical goal

of this project was to build a new device for archiv-

ing small physical objects in digital form on these ad-

vancing future platform for helping researchers con-

duct research. The proposed imaging device needed

to have speciﬁc features for this use and we designed

the system to satisfy this goal. Our technique can scan

3-dimensional structure, all-focused color image, dif-

fusion and transmission coefﬁcients of a target object.

A small object often has very ﬁne and minute

structures that require a sophisticated optical device

such as a microscope to be observed. For generating

an all-focused image, where all the picture elements

are focused, a con-focal microscope is one effective

choice to scan a target object in high-resolution and

high quality. However, it has a deﬁciency in that it

cannot capture color information. Traditionally, for

creating archives of small physical objects from ar-

chaeological sites,microscopic cameras and other 2-

dimensional optical apparatus have been used. This

kind of an image resource is a historical asset that

provides cultural and scientiﬁc explanation on what

our ancestors’ life and technologies of those days

were like. These replicable resources are one of the

most important elements in conducting an archaeo-

logical research. However, the imaging technique

is accompanied by loss of information that derives

from transformation of 3-dimensional structure to the

2-dimensional medium. To solve this problem, 3-

dimensional scanning technologies have been attract-

ing attention in the research ﬁeld. A 3-dimensional

scanner is a device that can capture 3-dimensional

structure of the target object often with color infor-

mation.

There are varieties of 3-dimensional scanners for

different uses. lazer-scanning technique is probably

the most popular technique and there are a number of

commercial products based on this technique. There

is another technique that utilizes an active lighting.

By projecting a certain pattern on a target object and

recognizing the projected pattern, this technique an-

alyzes the 3-dimensional surface shape of the target

object. In these two methods, color-information are

captured in a separate process often with a differ-

ent imaging device. In Stereo camera method, cor-

responding two points in two images taken by a set

of camera arranged in parallel are extracted to recon-

struct 3-dimensional surface structure of a target ob-

ject. This method is effective in scanning a large-scale

object such as an architectural structure and a land-

scape(Levoy, 2000). However, they have deﬁciency

in that they cannot deal with an extremely small ob-

ject less than 5 cm.

2 SYSTEM DESIGN

Our Micro Archiving technology can comprehen-

sively handle high-resolution all-focused imaging, 3-

dimensional capturing and transparency capturing si-

multaneously. Constructing high-resolution 3D mod-

els is an expensive and time-consuming task. In often

cases, 3-dimensional scanners and other image-based

modeling methods are used to automate this process.

However, existing systems had a trouble in case of a

small object. This is the main issue that our newly

proposed method solves. As we design the system,

we examined required features and functions that sat-

isfy our goal of providing researchers with an ultimate

archiving system.

2.1 Depth Measurement

It was the primal objective that our system can cap-

ture 3-dimensional structure of a small object that ex-

isting technique cannot deal with. Small objects, es-

pecially in case of a natural object and small animals

such as mineral materials and insects, have complex

and minute structures of ﬁne concavity and convexity.

Our system needed to capture these physical

features. Our methodology is based on an exist-

ing image-based model acquisition method called

”Shape-from-focus”(Nayer and Nakagawa, 1994). In

general, Shape-from-focus method is done in the fol-

lowing process. Firstly, a optical device, usually a

camera, is focused onto a part of the target object. By

slightly changing the focal length of the optical de-

vice, a sequence of images are taken each of these

images contains different region in focus as shown in

ﬁgure 2. In general, focusing depends on the distance

from the camera to the target object. Therefore, by

extracting focused pixels in each image, it is possi-

ble to measure the 3-dimensional surface shape of the

target object.

However, the drawback of this method is that it is

difﬁcult to generalize the focus model. Especially in

case of a complex object, pixels that are out of focus

affect neighboring pixels and this effect is extremely

complex to analyze. In general, a pixel becomes the

GRAPP 2007 - International Conference on Computer Graphics Theory and Applications

146

Figure 2: Shape from Focus.

brightest when it is in focus. The ﬁgure shows transi-

tion of brightness of a pixel in the captured sequence

of images. The x-axis represents the distance from the

camera to a part on the target object corresponding to

the pixel. As can be seen, it is difﬁcult to know at what

position the pixel is focused(Figure 3 : normal light-

ing). This is largely because of complicated blur ef-

fect of neighboring pixels. This effect becomes more

complex when measuring the 3-dimensional structure

of a very complex object.

Figure 3: Transition of Brightness under our slit-lighting

equipment... or a normal lighting equipment.

To simplify this complex blur effect, we designed

a special slit lighting system as shown in ﬁgure 4. The

lighting system and a digital camera is set up on a

shock prevention table. The slit lighting system is

composed of 8 illuminators which are attached to a

ring plate with a hole in the middle. A target object

goes through this hole when it is scanned. Each of the

illuminator is connected to a light-box through an op-

tical ﬁber cable. Light comes from the light-box and

gathered and bundled to about 0.5 mm thick by a lens

in the illuminator (Figure 5).

The slit-lighting system is arranged so that the

beam of light goes parallel with the focal surface of

the camera lens and it illuminates only the part of

the target object that is located very near to the fo-

cused part. Therefore, a sequence of captured images

contains mostly focused part radically eliminating the

complex blur effect.

Under the slit lighting system, the transition of the

brightness of each pixel becomes extremely simple

and it also becomes easy to ﬁnd focused pixels (Fig-

ure 3 : special lighting). We have not yet done an

elaborated evaluation of the effectivity of this lighting

system. However, with the proposed hardware pre-

operation utilizing the slit-lighting, a set of technical

problems in case of analyzing a structure of a small

complex object can be resolved.

As mentioned, our technique is based on Shape

from focus method. In our proposed method, a target

object is moved instead of changing the focal-length

of the camera lens. Our system has an object manip-

ulation arm that is attached on the shock prevention

table and moves a target object in high precision with

a built-in precision stepping motor. For a target ob-

ject in 2.0 cm to 5.0 cm length, the manipulation arm

is set to move in 0.1 mm pitch. For an object smaller

than 2.0 cm, it is set to move even in smaller pitch of

0.05 mm.

2.2 Resolution and Image Quality

To archive a target object in great detail, the capturing

device needs to be able to deal with high-resolution

imaging. As we used a high-end digital camera, the

resulting image can be as high-resolution as what the

image sensor can handle. However, in conventional

photography, it has been impossible to focus onto the

whole part of the target object at once because of tight

depth-of-ﬁeld especially in case of capturing a minute

object. This problem becomes even more critical in a

microscope. We solved this problem by utilizing the

depth data that is already captured by the system. To

capture the target object in natural color, the system

takes a sequence of pictures under a ﬂash lighting.

Since we already know what part is in what depth pa-

rameter meaning that it is possible to extract pixels in

focus from the image sequence.

The sequence of images is taken under 4 strobo-

scopic lighting devices (TWINKLE FII by COMET)

with a set of diffusers. With these lighting devices,

a target object is uniformly illuminated. As is men-

tioned in the following chapter, this lighting does not

necessarily have to be in this way and it can be any

kind of lighting condition as is used in various kinds

of existing active stereo 3D scanning methods.

This synthesizing technique makes it possible to

take images of the target object under various kinds of

lighting condition. Therefore, our method can be ex-

tended to support other image-based modeling tech-

nique. One of these modeling techniques is extraction

of reﬂection model. It is often that physical object is

composed of many different materials with different

physical characteristics. For example, a shell surface

of insects in a distinctive structure called structural

A VIRTUAL ENVIRONMENT FOR ARCHIVING MICRO-PRESENCE WITH IMAGE-BASED MODEL

ACQUISITION

147

Figure 4: Slit Lighting.

Figure 5: Slit Lighting structure.

Figure 6: Transition of Brightness under our slit-lighting

equipment.

Figure 7: Transition of Brightness under our slit-lighting

equipment.

color. The surface emits various colors depending on

the lighting condition. Being able to manipulate the

lighting condition allows modeling this kind of reﬂec-

tion pattern. Another example is transparency of the

target material. We in fact designed a lighting system

and transparency scanning technique using it to cap-

ture diffusion and transmission coefﬁcients. This is

discussed in the following chapter.

We used a high-end consumer digital camera

(Canon EOS-1Ds Mark II) and SIGMA MACRO 50

mm F2.8 EX DG lens. The camera resolution is 5000

x 3400 pixels and the resulting depth data is also in

this resolution, which is almost unattainable in exist-

ing 3-dimensional scanners. Our system does not de-

pend on a speciﬁc function of a certain digital camera.

The current camera can be replaced by any kind of a

new product with higher resolution imaging capabil-

ity.

2.3 Capturing Other Physical Features

The lighting condition a sequence of images for gen-

erating all-focused image can be in any form for mea-

suring other physical features. By designing a pro-

prietary background illuminator, our system can also

capture transparency of the target object. This is an

element that did not exist in the conventional photog-

raphy.

In the natural world, there are semi-transparent

materials such as a glass, feather of insects. To

represent these semi-transparent objects in com-

puter graphics, it is indispensable to take into ac-

count transmission, refraction and diffusive reﬂection.

To deal with these optical characteristics of semi-

transparency in photography, it is needed to check

the materiality of the object corresponding to each

pixel of an image. However, it is time-consuming,

expensive and practically difﬁcult. In existing re-

search, there are several methodologies proposed that

uses blue-screen to measure semi-transparency. In

fact, even with these methods, semi-transparency is

assigned to a virtual object in computer graphics by

empirical assessment of an engineer and an artist.

Also, diffusion effect, where a background image gets

blurred in accordance with the distance between the

background object and a foreground semi-transparent

object, is hardly considered. In case of need, an ob-

ject is modeled so that it appears to have a diffusion

effect. In this research, we propose a new methodol-

ogy to capture two characteristics of transmission and

diffusion effect of a thin semi-transparent object from

a series of photographic images and represent them in

computer graphics.

In general, transparency means the ratio of how

GRAPP 2007 - International Conference on Computer Graphics Theory and Applications

148

a foreground object is mixed with a background ob-

ject. Diffusivity means how the background object

gets blurred as the distance between the foreground

object and the background object increases. In the

current state of the industry, assigning these parame-

ters to 3-dimensional models for use in ﬁlms and com-

puter graphics is empirically done and never seriously

considered. We succeeded in developing a system to

capture these parameters from a real object.

On top of this, the resulting scanned model is in

general 3D computer graphics format. Therefore, it

can be edited on various kinds of commercial 3DCG

software and also shown on general viewers such as

a regular 3D viewer software and Web3D viewer for

observation. Using a general format allows other

structural features such as internal structure captured

by X-ray inspector to be easily added to the captured

model.

3 SCANNING PROCESS

Our Micro Archiving scanning process is mainly

composed of three parts in shown ﬁgure 8: scanning,

image processing and postprocessing. In the scanning

process, sequences of images are captured by our pro-

posed system. In the image processing, the sequences

of images are processed to extract 3-dimensional sur-

face shape, an all-focused image and a transparency

map. Finally, in the post-processing, these data are

EOS

Figure 8: Transition of Brightness under our slit-lighting

equipment.

combined together to output a 360 degree model and

texture data. Also the model data is converted into a

generic 3D model data format and also Web3D format

for rendering.

3.1 Initialization

Firstly the target object is set on the moving arm and

all the mechanical parts are set to be at the initial po-

sition. The slitlighting is set to illuminate the top of

the target object. The lens of the digital camera is also

set to focus on the illuminated part.

Camera parameter is also critical and it affects the

quality of the resulting model data. It needs to be ﬁne-

tuned depending on what the target object is. How-

ever, we heuristically know that the following param-

eters are the best as default settings for an average

target object.

Table 1: Camera parameters.

For slit-lighting condition

Exposure 1.00 [sec]

Lens Aperture F/5.6

ISO 400

Shutter Speed 1.00 [sec]

Aperture F5.6

For strobo lighting condition

Exposure 1/160 [sec]

Lens Aperture F/8.0

ISO 100

Shutter Speed 1/160 [sec]

Aperture F8.0

After the initial calibration, all the steps of captur-

ing and processing are done by an automated script

program.

3.2 High-Resolution 3-dimensional

Imaging

Because focusing depends on the distance from the

camera to the corresponding part in the image, it is

possible to calculate how far from the camera each

pixel in a taken image is.

Img(x, y) : a pixel at the coordinate(x, y)

|Img(x, y)| : brightness of the pixel at the coordinate (x,y)

ImgAF : all-focused image

ImgD : depth image

ImgS

Index

: Index

image taken under the slit lighting.

ImgC

Index

: Index

image taken under the ﬂash lighting.

For measuring 3D structure, the following process

is automatically done.

|ImgD(x, y)| = Index

where,

|ImgS

Index

(x, y)| =

max(|ImgS

(x, y)|, |ImgS

(x, y)|, ··· , |ImgS

(x, y)|)

This process is done by our proprietary software

called MA Synthesize.

A VIRTUAL ENVIRONMENT FOR ARCHIVING MICRO-PRESENCE WITH IMAGE-BASED MODEL

ACQUISITION

149

3.3 High-Resolution All-Focused

Imaging

For generating an all-focused image, the following

process is automatically done. For this process, depth

image that is acquired in the previous process is used

to specify the pixels used to construct the all-focused

image.

ImgAF(x, y) = ImgC

|ImgD(x,y)|

(x, y)

Our proprietary computer program synthesizes

these images to put together only focused pixels to

construct an allfocused image.

3.4 Transparency Capturing

In the natural world, there are semi-transparent mate-

rials such as a glass, feather of insects. To represent

these semitransparent objects in computer graphics,

it is indispensable to take into account transmission,

refraction and diffusive reﬂection(Figure 9).

Figure 9: Diffusive Effect.

To deal with these optical characteristics of semi-

transparency in photography, it is needed to check

the materiality of the object corresponding to each

pixel of an image. However, it is time-consuming,

expensive and practically difﬁcult. In existing re-

search, there are several methodologies proposed that

uses blue-screen to measure semi-transparency. In

i-3

i-2

i-1

i+1

i+2

i+3

i-3

i-2

i-1

i+1

i+2

i+3

Background Lighting

Foreground Lighting

Figure 10: Lighting Model.

fact, even with these methods, semitransparency is

assigned to a virtual object in computer graphics by

empirical assessment of an engineer and an artist.

Also, diffusion effect, where a background image gets

blurred in accordance with the distance between the

background object and a foreground semi-transparent

object, is hardly considered. In case of need, an ob-

ject is modeled so that it appears to have a diffusion

effect. In this research, we propose a new methodol-

ogy to capture two characteristics of transmission and

diffusion effect of a thin semi-transparent object from

a series of photographic images and represent them in

computer graphics.

The target object is ﬁxed between the vertical

lighting system and the background illuminator. As

shown in ﬁgure 10, assuming that the vertical light-

ing is composed of a number of points U

and the

background illuminator is composed of points Q

, the

color Γ(x) = [x

, x

]

which is observed at the

point x on the camerafs projection plane can be de-

scribed by Lambertfs cosine law and inverted square-

root law as follows.

γ(x) = κ

′

∞

−∞

∞

−∞

cos



arctan



′2

′



′2

+ Z

′2

′

κL

∞

−∞

∞

−∞

cos



arctan





+ Z

+ ιL

= 2πκ

′

+ 2πκL

+ ιL

(1)

Here, U

− P = [X

′

, Y

′

, Z

′

]

− P =

, Y

, Z

]

′

are diffusive reﬂection coefﬁcients

when observed from the front side, κ is the diffusive

reﬂection coefﬁcient when the semitransparent object

is observed from the backside and κ represents trans-

mission coefﬁcient of the semi-transparent object

observed from the backside. It is assumed that the

brightness U

of the vertical lighting U

is composed

of uniform color elements L

= [L

, L

]

and

the brightness Q

of the background illuminator

is also composed of uniform color elements

= [L

, L

]

Estimation of the color of semi-transparent object

Γ(P) and its diffusive reﬂection coefﬁcient a The sum

of κ and ι is deﬁned to be diffusive reﬂection coefﬁ-

cient a.

Using α, the equation (1) can be rewritten as fol-

lows.

Γ(x) = (1− α)Γ(P) + αL

, Γ(P) =

2πκ

′

1− α

(2)

Here, the point Γ(P) can be presumed to be a point

on the semi-transparent object P.

From the equation (2), By measuring the color

of the semi-transparent object Γ(x

) captured on a

white background, the color of the white background

, the color of the semi-transparent object Γ(x

)

captured on a gray background and the color of the

gray background L

, it is possible to calculate Γ(P)

and α shown in the table 2.

GRAPP 2007 - International Conference on Computer Graphics Theory and Applications

150

Estimation of diffusive reﬂection coefﬁcient κ and

transmission coefﬁcient ι, acquired α split into diffu-

sive reﬂection coefﬁcient κ and transmission coefﬁ-

cient ι.

With an object that has a known diffusive reﬂec-

tion coefﬁcient, it is possible to measure κ

′

of a semi-

transparent object. In our method, a target object is

assumed to be a very thin object. Therefore, it is as-

sumed that κ can be approximated to κ

′

as long as it

does not exceed α and κ, ι are approximated by the

following equation.

′

= min(α, κ), ι = α − κ

(3)

Acquired κ, ι are shown in the table 2.

Table 2: Acquired transparency maps of a sheet of japanese

hand-made paper.

Img(L

) Img( Γ(x

)) Img(L

) Img(Γ(x

))

Img(Γ(P)) Img(α) Img(κ) Img(ι)

3.5 Rendering

To render with acquired coefﬁcients, three images,

the rendered image of the semi-transparent object

Img( Γ(P)), the image blurred in accordance with the

distance between the background object and the semi-

transparent object Img(Blur(L

)) and the background

image Img(L

), are prepared. As shown in the table

2, Img(α), Img(κ), Img(ι) are applied to the blending

map of each image.

We proposed a method for measuring diffusive re-

ﬂection coefﬁcient and transmission coefﬁcient from

photographic images and a rendering method for

them. By this, it became possible to easily mea-

sure pixel-by-pixel transmission and diffusion effect

from a photography. It also became possible to repre-

sent diffusion effect, which has been hardly consid-

ered in conventional computer graphics. Since we

can not exactly know diffusive reﬂection coefﬁcient

and transmission coefﬁcient, which this paper deals

with yet, the validity of our method has not yet been

fully proved. However, because elements needed for

photo-realistic computer graphics are variables rela-

tive to neighboring pixels, our method is expected to

be highly effective.

transparency map

all-focused image

rendered images

Figure 11: Examples of scanned transparency map and ren-

dered images.

4 POST PROCESS

4.1 Integrating Coordinate System

Our system scans a target object from multiple direc-

tions. The multiple model surfaces are in different

coordinate systems and needed to be in a common

coordinate system to be aligned and merged. Our

technique does this by calculating how big in a cap-

tured image an object in a known size is represented.

This size is represented in a unit called PPM(pixels

per millimeter). The XY coordinates of the scanned

all-focused image and the Z coordinate of the depth

data are converted into a common 3-dimensional co-

ordinate by this parameter.

4.2 Mesh Alignment

Because our system can capture a target object from

only one direction at one time. To create a model that

can be observed from 360 degree, mesh data from

multiple direction have to be combined seamlessly.

Our system utilizes a commercial software called Ge-

omagic Studio 6. This software is used to align sur-

face models captured from different angles.

4.3 Mesh Merging

Once the surface models are aligned, they are merged

together to form a continuous surface. This opera-

A VIRTUAL ENVIRONMENT FOR ARCHIVING MICRO-PRESENCE WITH IMAGE-BASED MODEL

ACQUISITION

151

tion is done by our proprietary software called Micro

Archiving Mesh Editor.

Figure 12: Mesh aligning.

5 RESULTS

The current maximum scannable texture size is 5000

x 3400 pixels and our technique allows maximum

model data size of about 30,000,000 bytes (2.3 G

bytes).

For a target object in the size of 2.0 cm long, it

takes about 30 minutes to scan from one direction.

Usually to create 360 degree model, a target object

needs to be scanned from 8 directions. Therefore it

takes 240 minutes to scan the whole target object.

6 APPLICATIONS

6.1 Web3d/webstereo

On the current broadband Internet, Web3D technol-

ogy can potentially be used for various kinds of

applications such as online museums and multime-

dia database systems for both general public and re-

searchers specialized in a certain region. This fact

will necessitate rapid development of content for such

applications. Our Micro Archiving technology pro-

vides a technical solution. We created some Web3D

contents public on an Internet website where any-

body can observe and interact with high-resolution 3-

dimensional models. Users can freely zoom into a

detail to observe the minute structures.

Figure 13: Web3D contents.

6.2 Animation

The 3-dimensional models captured by our method

can be edited on many kinds of 3D computer software

for various purposes. One possible application would

be 3D computer animations. Such as digital ﬁlms,

games and other many different kinds of entertain-

ment media that requires photorealistic 3-dimensional

models.

Figure 14: Animation sequence.

6.3 Academic Research

Entomology, which is a study on insects, often re-

quires real specimens of insects to conduct a research.

The models that can be captured by our methodol-

ogy has enough resolution, accuracy and dense infor-

mation including color, shape and transparency. We

also succeeded in combining X-ray scanning struc-

tural data. With our proprietary interaction software,

anybody can interact with and observe the 3D models.

This would be strong application software for aca-

demic researchers who always need access to speci-

mens.

Figure 15: A proprietary interactive viewer.

REFERENCES

Levoy, M. (2000). The digital michaelangelo: 3d scanning

of large statues. In Proceedings ACM SIGGRAPH

2000. ACM.

Nayer, S. K. and Nakagawa, Y. (1994). Shape from fo-

cus. In IEEE Transactions on Pattern Analysis and

Machine Intelligence. IEEE Computer Society.

GRAPP 2007 - International Conference on Computer Graphics Theory and Applications

152