VIRTUAL RESTORATION OF A MEDIEVAL POLYCHROME

SCULPTURE

Experimentation, Modelization, Validation and

Visualization in Spectral Ray-tracing

Sylvain Dumazet, Patrick Callet

Applied Mathematics and Systems Lab., Ecole Centrale Paris, grande voie des vignes, Chˆatenay-Malabry, France

Ariane Genty

INOLAM, 26 bis rue Kl´eber, Montreuil, France

Keywords:

Ray-tracing, spectrophotometry, 4-ﬂux model, sculpture, virtual restoration, gilding, painting.

Abstract:

A pluridisciplinary work always in progress involving 3D digitization, simulation, rapid prototyping, virtual

restoration of a french medieval sculpture is presented. This work is led in the framework of a general

collaboration between three academic labs, industrial partners and Cultural institutions. The main purpose is

to virtually represent a polychrome statue of the XIIIth century in high quality spectral rendering, to simulate

its visual and original appearance at that period. The complete process used throughout all the phases of the

project mainly involves optical devices that ensure no physical contact with the museum object.

This article describes the complete chain of engineering resources and the main models we used for

accomplishing our objective. From 3D capture without contact to plaster replica, the complete process will be

described and illustrated with images and objects during the conference. Some sequences extracted from the

didactic and scientiﬁc movies produced will also be presented.

1 INTRODUCTION

A collaboration with the ”Centre des Monuments Na-

tionaux” (CMN) helps us to elaborate a new project:

the study of a polychrome medieval statue, the re-

cumbent statue of Philippe Dagobert de France (circa

1222 - 1232 AC) in the Saint-Denis Basilica(Fig. 1),

the royal necropolis in the north of Paris. The CMN

being in charge of organizing and managing the pub-

lic visits in more than one hundred French monu-

ments, is deeply involved in the realisation of the

study. The research results are meant to be shown,

permanently near the Philippe de France tomb, in

order to raise a large public awareness to the sci-

entiﬁc research contribution, and for a better under-

standing of historical heritage. The presented work

takes place in a previously deﬁned general frame-

work known ”OCRE method” (Optical Constants for

Rendering Evaluation) (Callet, 1998), developed by

Patrick Callet and describing the optical behaviour of

materials on the basis of their fundamental properties.

For metals and alloys we naturally used their com-

plex indices of refraction, either computed or care-

fully measured, but, in any case always validated by

measurements. Plasma physics and spectroscopic el-

lipsometry, for extracting real and imaginary parts of

the complex indices of refraction, were used for that

Figure 1: The recumbent statue of Philippe Dagobert in

Saint-Denis Basilica (13th century AC).

463

Dumazet S., Callet P. and Genty A. (2008).

VIRTUAL RESTORATION OF A MEDIEVAL POLYCHROME SCULPTURE - Experimentation, Modelization, Validation and Visualization in Spectral

Ray-tracing.

In Proceedings of the Third International Conference on Computer Graphics Theory and Applications, pages 463-470

DOI: 10.5220/0001100204630470

 SciTePress

pertinent modelling. These previous important results

are very useful for the rendering of the golden parts

of the statue that consist in a gold alloy containing

a small amount of silver. The metallurgical compo-

sition of the gold leaves recovering the visible rests

will greatly help in the formulation of the visual ap-

pearance for virtual restoration or historical reconsti-

tution. The last study about the visual inﬂuence of

the underlaying paint, called ”bole”, on a gilded sur-

face has shown the importance of the transparency of

a thin metallic gold ﬁlm and of the associated small

cracks due to the application process (called ”burnish-

ing”)(Dumazet et al., 2007).

We here focus on modelling and representing the

painting materials using spectrophotometry and ex-

trinsic physical parameters such as the paint ﬁlm

thickness, the concentration of each species of pig-

ments, their mean diameter or the granulometry dis-

tribution functions,... Intrinsic parameters, charac-

terizing the nature more than the structure of all the

compounds such as complex indices of refraction of

all involved materials are used throughout the study.

As an extension of the Kubelka and Munk theory,

a four-ﬂuxes approach of the multiple scattering of

light is used for the simulation. At each step of the

study we compare our computed results of simula-

tion with measurements made on handcrafted sam-

ples in laboratory (Fig. ﬁg:peinture-atelier) or in situ,

i.e. on the statue itself. Helped by the ”Centre des

Monuments Nationaux” (National Monuments Cen-

tre), and with the knowledge about painted stones

during the middle-age, we studied the pigments and

binders which were probably used by the artist. In

a ultimate step we shall replicate the statue to 1/3rd

of its original dimensions with its most probable and

original colours.

2 HISTORICAL CONTEXT

This recumbent statue has been chosen for its re-

maining polychrome traces (Fig. 1) and also because

it was representative of the ﬁne middle XIIIth cen-

tury funeral sculpture. At that time, Louis IX, not

yet known as king ” Saint Louis”, was setting up

the Saint Denis Royal necropolis and the ”Children

of France” necropolis in the Royaumont Cistercian

abbey at Asni`eres sur Oise (Erlande-Brandenburg,

1975). It concerns the Philippe-Dagobert’s tomb, a

young Saint-Louis’s brother born in 1222, who was

designated to be ”clericus”, i.e. to have an ecclesiastic

career, died in 1232 and buried in Royaumont (North

of Paris). His tomb has been raised in the abbey-

church chancel between October 1235 (consecration

Figure 2: Philippe Dagobert - (left) Stained glass window

drawing, (right) Drawing of the tomb with its recess.

date of this church) and middle XIIIth. He is not rep-

resented as a child here but as an idealised young man,

laying down, opened eyes, joint hands; his head lay

on two cushions each of them being held by two an-

gels; his feet are resting upon a lying lion holding

in its paws a leg of deer. With the French revolu-

tion this tomb encountered several dramatic events.

It has been transferred in Saint-Denis in 1791, then

damaged in 1793-94 and deserted till 1796 when,

it has been preserved in the Mus

ee des monuments

franc¸ais. In 1816 it returns to Saint-Denis, where it

has been installed in the crypt; then the famous archi-

tect Viollet-le-Duc, between 1860 and 1867, inspired

by the original pieces, rebuilt entirely the sarcopha-

gus and installed it in the north transept. Concerning

the paintings, Baron de Guilhermy published his ob-

servations in 1848 (de Guilhermy, 1848) : ”From the

restitution of these tombs in Saint-Denis, the whole

sculpture has been restored, and the old paintings

disappeared under a new covering which after only

thirty years were already eroded”. It refers to painted

work ordered by the architect Franc¸ois Debr´e in 1820.

Luckily, half a century ago, Millin could study and

draw the Philippe-Dagobert tomb in Royaumont as it

was in 1790, just before it was transferred to Saint-

Denis (Millin, 1791). Millin most likely describes

the remnants of medieval paintings namely blue for

the cappa magna sprinkled with golden squares and

diamonds. This pattern is conﬁrmed by a drawing

made in 1694 for Roger de Gaigni`eres, which rep-

resents a missing stained glass window in the Roy-

aumont abbey-church with the young prince up and

wearing the same clothes than his recumbent statue

(left ﬁgure 2). As for the tomb, Gaigni`eres made

it drawn and painted with great precision. In the right

part of ﬁgure 2 is exhibited a reproduction of the orig-

inal coloured drawing conserved in the Bodleian Ox-

ford Library. This research could allow to show once

again how colour was present at the medieval epoch.

As attested on the picture (Fig. 1) there are still some

traces of paint and gilding on the stone at the contrary

of the other graves gathered in the basilica. The robe

is blue with red sleeves, he has yellow hair, and he

GRAPP 2008 - International Conference on Computer Graphics Theory and Applications

464

is wearing black breeches. His head rests on a red

cushion on top of a green one. The angels’bodies are

blue painted and their wings were gilded during the

last restorations in XVIII and XIXth centuries. For

discriminating all the superimposed layers of paints

it was decided to make some takings for chemical

analyses.

3 SCIENTIFIC METHODOLOGY

AND TECHNICAL PROCESSES

Two kinds of data are necessary for the visual recon-

struction of the statue. Shape and spectrophotometric

informations are required. The ﬁrst one is more fre-

quently used in computer graphics while the second

is generally replaced by a trichromatic description of

the actual colors.

3.1 3D Shape Acquisition and Virtual

Reconstruction

As for previous projects the 3D digitization has been

performed in situ (in Saint-Denis Basilica) after pub-

lic hours, without contact using an optical system

based on structured light projector and a camera (Fig.

3) from Breukmann corp. A regular light and shadow

grid is projected on the object and the distortion of the

boundaries between light and shadows on the surface

is recordedby the camera and analysed, in order to ex-

tract a cloud of 3D points with a good accuracy. This

gave us 167 clouds of 3D points used for reconstruct-

ing the surface with RapidForm, a software from Inus

Technology corp.

The ﬁnal 3D shape of the tomb (recumbent statue

on the sarcophagus) in full resolution represents about

7,5 Million triangles (Fig. 3).

Figure 3: (left) 3D digitization using structured natural

white light. (right) 3D colorless model displayed with Catia

(Dassault System software).

Figure 4: 3D model cut up not completely decomposed in

color parts.

3.2 Spectrophotometric Data

Acquisition

The portable spectrophotometer (USB 2000 series

from Ocean Optics corp.) consists in a set of opti-

cal ﬁbers guiding a calibrated light source (a halo-

gen lamp for CIED65 illuminant) surrounding the

backscattered light guiding ﬁber. The latest is con-

nected to a spectrometer where a 1024 photodiodes

array transforms the received light in a digital sig-

nal. We then obtain the diffuse reﬂectance factor of

the paint according to the visible wavelengths domain

(from 380nm to 780nm). Also in situ, we captured

spectral information on the remaining traces of paint

thanks to a spectrophotometer with optical ﬁber use-

ful for further analyses and material characterization.

The portable spectrophotometer and the recorded

refectance factors are shown in Fig. 5.

As we need to compare and validate permanently

our choices and models, we also used pigments and

binder samples prepared in the laboratory. We also

made our spectral measurements on these samples

(Fig. 5).

Figure 5: Spectral data acquisition in situ, reﬂectance fac-

tors layout and analysis.

3.3 Painting Materials

During the Middle Age and more particularly during

the thirteenth century, we already know, (Escalopier,

2004) what pigments are the most commonly used

and also how the paints were applied on the sculp-

ture. First, the artists start with two or three layers

of ceruse, made out of water and white lead. This

VIRTUAL RESTORATION OF A MEDIEVAL POLYCHROME SCULPTURE - Experimentation, Modelization,

Validation and Visualization in Spectral Ray-tracing

465

Figure 6: Medieval paint samples preparation.

step enables to waterproof the stone but also enables

the painters to rectify the surface defects of the stone

by coating. Then they apply several layers of paint,

consisting of pigments embedded in a binder like

egg (tempera technique), animal protein or gum-resin.

The most frequently encountered pigments are:

• White : White Lead, White Chalk, Lime;

• Black : Wood Smoke, Black Vine;

• Blue : Azurite, Lapis-Lazuli, Indigo;

• Red : Vermilion, Red Lead, Red Ochre;

• Yellow : Yellow Ochre, Massicot;

• Green : Green Clay, Malachite, Verdigris.

Which of them were used in the speciﬁc case of

Philippe Dagobert’s recombent statue? The recorded

diffuse reﬂectance factors, obtained by the only pos-

sible analysis involving a non destructive method al-

low us to compare them to the results obtained by the

chemical analysis for validation and simulation.

We also need to specify which part of the global re-

constructed shape would be associated with each kind

of materials (pigments and binder or/and gilts) (Fig.

4).

3.4 The Kubelka-Munk Model

Extended to Four-ﬂuxes Theory

Other works are related to spectral representation of

colours (Sun et al., 2000), (Rougeron and Per-

oche, 1997), (Devlin et al., 2002), (Gondek et al.,

1994), and pigmented materials in computer graph-

ics (Haase and Meyer, 1992), (Baxter et al., 2004),

using the Kubelka-Munk theory. Our ﬁrst concern

here was, to use an extended Kubelka-Munk model

to four ﬂuxes approach to better ﬁt a modeling of

paintings and to realize our calculations based on 81

wavelengths bands of 5nm over the visible spectrum

[380;780]nm. Our second concern is in our entire

process to constantly go back and forth between our

theoretical models and our measurements on real ma-

terials. And above all, helped with our collaborations

to apply these concepts in an archeological, and his-

torical approach. The Kubelka-Munk model is almost

well suited for the description of pigmented materi-

als. These are described as a scattering medium, lay-

ing on a scattering background. The system is illu-

minated by a diffuse orthotropic incident light. It can

be demonstrated (Volz and Teague, 2001; Callet and

Zymla, 2004), that an incident orthotropic light ﬂux

on such a ﬁlm of thickness h can be considered equiv-

alent to a collimated directionnal and normal incident

ﬂux on a paint ﬁlm of thickness 2h. The model gives

the reﬂected and the transmitted ﬂuxes from an inci-

dent light, normal to the layers across a paint ﬁlm of

thickness h laid on a substrate. The plain Kubelka-

Munk theory (Kubelka and Munk, 1931) is then re-

duced to an equivalent 2-ﬂux theory involving two di-

rectional ﬂuxes of light and, according to the previ-

ous remark above, one going downward L

and the

other upward L

−

. The layer of paint is a macroscopic

scattering and an absorbing medium so we consider

two coefﬁcients: S and K the scattering and absorp-

tion ones. These last coefﬁcients depend on the pig-

ments diameters and consequently on the granulome-

try distribution function. A ﬁne grinding involves an

important multiple scattering inside the paint ﬁlm and

gives a desaturated colour obtained without any addi-

tion of white pigments. This is a physical whitening

only. Further works will use the micro-stratigraphy

images for extracting a best estimation of the granu-

lometry distribution function. All the involved terms

are wavelength dependent. The substrate has a re-

ﬂectance factor R

. Thus, we account for the nor-

mal and directional lighting on the outer layer. So the

2-ﬂuxes model is improved with 2 additional ﬂuxes

Figure 7: Four-Fluxes theory: the directional (l

−

) and

diffuse (L

−

) ﬂuxes, together with the boundary condi-

tions.

GRAPP 2008 - International Conference on Computer Graphics Theory and Applications

466

of light l

−

and l

normal to the interfaces. Includ-

ing the two previous diffuse ﬂuxes, one downward

and the other upward L

−

(as described in (Volz

and Teague, 2001)). Figure 7 shows the principle of

this model. The incident light is then decomposed

in a remaining directional reﬂected light and of an

additional scattered light due to volume and surface

scattering from the substrate. Two specular compo-

nents are then added to the classical Kubelka-Munk

model. We operate a local radiative balance and write

four equations, where k

′

, is the absorption coefﬁcient

for directional light, and respectively s

, the forward

scattering coefﬁcient, and s

−

, the backward scattering

coefﬁcient.

= −(k+ s

+ s

−

dz (1)

−dl

−

= −(k+ s

+ s

−

dz (2)

= s

dz+ s

−

dz− (K + S)L

dz+ SL

−

(3)

−dL

−

= s

−

dz+ s

−

dz− (K + S)L

−

dz+ SL

(4)

Setting l

and l

−

equal to zero and solving the

above system of equations, leads to the plain 2-ﬂuxes

Kubelka and Munk expressions of absorption coef-

ﬁcient K and scattering coefﬁcient S. The measure-

ments leads to :

• h: the layer thickness ;

• R

: the background reﬂectance ;

• R

∞

: the reﬂectance for an inﬁnite thickness (to-

tally opaque layer) ;

• R: the layer reﬂectance (what we need).

We successivelycalculate: R

, the surface reﬂectance:

∞

− R)

− R

∞

(1− R

∞

+ R

(5)

S, K the macroscopic scattering and absorption coef-

ﬁcients:

S =

2.303

∞

1− R

∞

log

∞

(1− R

∞

)

∞

− R

(6)

K =

2.303

1− R

∞

1+ R

∞

log

∞

(1− R

∞

)

∞

− R

(7)

The determination of the paint characteristic ﬁlm

thickness h is made using the micro-stratigraphic im-

ages (Fig. 8). For the four-ﬂuxes model we add the

speciﬁcation of all the terms k, s

, s

from the optical

coefﬁcients. Let :

a = 1+

, b =



+ 2



x = bhS, A = asinh(x) + bcosh(x)

Figure 8: Microstratigraphy of a paint scale taken on the

green bottom cushion. Each successive layer gives infor-

mation about its mean thickness and the granulometric dis-

tribution of the embedded pigments. Magniﬁcation x20.

from which we deduce the following parameters

τ =

, ρ

= a − b, ρ

sinh(x)

We obtain from the previous relationships:

= e

−µh



r,0





∗

r,0



Then:

p =

∞

(1− τT

) − R

(1− τT

) − ρ

q = ρ

p− R

∞

Next, inverting p and q equations leads to the scatter-

ing coefﬁcients:



= p(µ− aS) + qS

= pS− q(µ+ aS)



where µ =

Finally we can deduce the absorption coefﬁcient for a

very thin layer:

k = µ− (s

+ s

)

These results were specially accurate for gilts: gold

leaf polished above a colored paint ﬁlm (the bole).

The very small holes observed inside the gold leaf en-

rich the reﬂected spectrum with the diffuse compo-

nent of the underlying paint. An other component of

diffuse reﬂectance is due to the small cracks produced

by the polishing process of the gold leaf. That inﬂu-

ence has been recently shown(Dumazet et al., 2007).

3.5 Diffuse Reﬂectance Factors and

Paints

We analyzed some paint samples made in laboratory

(Fig. 6) and worked on the appearance of the paints

VIRTUAL RESTORATION OF A MEDIEVAL POLYCHROME SCULPTURE - Experimentation, Modelization,

Validation and Visualization in Spectral Ray-tracing

467

Figure 9: Reﬂectance factors of the blue pigments, azurite

and lapis-lazuli depending on the paint ﬁlm thickness.

depending on the thickness of each deposited ﬁlm.

We also measured the thickness of each sample. First,

we notice the change in reﬂectance as the thickness

increases. The thicker the layer the lower the re-

ﬂectance. This means that the shade is darker as the

thickness is increased, thus masking more and more

the white substrate. For several pigments (green clay,

malachite, black vine, vermilion, and lemon ochre),

the reﬂectance factor for two and three layers tend to

be quite similar, so we will consider that the diffuse

reﬂectance factor for the three layers sample gives the

term R

∞

to determine the coefﬁcients K and S of these

pigments. Spectral measurements help in differenti-

ating a lapis-lazuli from an azurite ﬁlm (Fig. 9). The

ﬁrst one has two peaks, one in the blue wavelengths

near 455nm and the other in the red and the near IR

wavelengths region (780nm). Only one peak appears

for the azurite pigments. Lapis-lazuli and its subtle

reddish component is known as natural ultra-marine

blue, very efﬁcient for reﬂecting infra-red radiations

(ideal paint used for the shutters in the mediterranean

countries). We can also notice that the two pigments

have different maxima in the blue region of the vis-

ible spectrum. One exhibits a relatively narrow peak

while an other has a more ﬂat and extended peak tend-

ing to cover the green shades region. Sometimes, a

small amount of malachite in natural azurite can ex-

plain the reﬂectance curve aspect and also the green-

ish color observed on the sample. Now that we have

the reﬂectance factors of the pure pigments, we want

to compare them to the ones we measured at the basil-

ica on the recumbent statue. We have identiﬁed some

of the pigments used on the statue. It seems that the

pigment used by the artist was identiﬁed as green clay,

according to the reﬂectance curves. We have made

this kind of comparison for each in situ reﬂectance

factor. The blue areas seem to have been painted with

azurite. The reﬂectance factor of the red cushion ap-

proaches that of the red lead. We can notice that the

Figure 10: Comparison of the in situ measured diffuse re-

ﬂectance factors with those of their corresponding hand-

crafted samples.

recorded spectra in the basilica have a weaker ampli-

tude between their maxima and minima than those of

the samples. This phenomenon expresses the fact that

the colors of the sculpture are now very de-saturated.

There are several reasons for that. First, the paint lay-

ers are old and dirty. Then, the deposited dust on the

sculpture surface increases the whitening by surface

scattering of all incident light.

More studies were necessary to exactly determine

the composition of each paint used to achieve this

sculpture. More samples of different pigments have

been elaborated. Therefore we virtually created mix-

tures of pigments and compared them to the in situ

recorded reﬂectance factors, according to the robust

color matching methods.

Figure 11: First results obtained with only spectropho-

tometric data and physical samples preparation. (left)

Philippe-Dagobert’s cappa magna rendered in lapis-lazuli

blue pigment by the radiosity software Candelux. (right)

Philippe-Dagobert’s sleeves - vermilion pigment rendered

with Candelux, opaque gold leaf covered lion rendered in

ray-tracing with Virtuelium. CIE D65 illuminants.

GRAPP 2008 - International Conference on Computer Graphics Theory and Applications

468

3.6 Spectral Simulations

With a correct characterisation of all materials and

their state of surface (waviness, roughness, paint-

ﬁlm thickness), the lighting conditions and a stan-

dard colorimetric observer (CIE 1964 10

◦

), we can

compute the restored visual aspect of the museum ar-

tifacts(Pitzalis et al., 2007). For paints simulations,

including thin or thick metallic reﬂection the im-

ages are computed with Virtuelium our multithreaded

ray-tracing software running generally over 81 wave-

lengths bands of 5 nm width, polarization of light,

CIE standard illuminants and colorimetric observer.

The 3D coordinates of the points are distributed over

an octree structure, made to speed up the ray-surface

intersections computations. We use the complex in-

dices of refraction which characterise the intrinsic

properties of all homogeneous materials. Optical

constants are then, the real part and the imaginary

part of the complex indices of refraction for the met-

als or alloys. These can be measured on real pol-

ished plates by spectroscopic ellipsometry, an optical

method based on Fresnel formulas and the analysis of

the amount of the reﬂected and polarized light at large

incidence on a smooth surface. The whole theory of

reﬂection of light by a metallic surface is available in

optics bookssuch as Born and Wolf’s (Born and Wolf,

1975) and for computer graphics and lighting in (Cal-

let, 2006; Callet, 2007) including quantitative data.

According to the physical parameters extracted from

microstratigraphy and spectrophotometric measure-

ments, we have computed with the 4-ﬂuxesmodel and

Virtuelium, the image presented in Fig. 12. Thus, the

successive layers with their thickness and concentra-

tion as plausible for the XIIIth century can appear. We

have made many simulations with different standard

illuminants and with characterized real light sources.

Some parts of the sculpture as the prince’s face or

the angel head were probably painted with a cinnabar

and vermilion mixture. No visible traces permitted to

conﬁrm that point and we decided to render the cor-

responding parts with the only white lead in a totally

opaquelayer. We computed the virtual restored aspect

of the medievalsculpture with the following materials

• angel tunic: lapis-lazuli layer on azurite ;

• angel wings: 50 % red lead and 50 % cinnabar ;

• Philippe Dagobert hair: 200 nm thick gold leaf on

yellow ochre substrate ;

• upper cushion: 50 % red lead and 50 % cinnabar ;

• lower cushion: not completely opaque layer of

malachite pigments.

Figure 12: The actual most plausible colors of the medieval

polychrome recumbent statue of Philippe Dagobert. Ren-

dering in spectral ray-tracing by Virtuelium on an INTEL

Pentium4 3GHz dual processors computer in 8 hours for

a 1200x1200 resolution, 1Gb RAM and running under the

operating system GNU/Linux - Fedora 5.

4 CONCLUSIONS

We went one step ahead in our knowledge and know-

how of cultural heritage engineering and physical

models for materials rendering in optical simulation.

Till now we worked with standard illuminants such

as CIE A, D65 or E illuminants. We shall need for

these studies a better knowledge on natural lighting

by stained glass windows and all other anthropogenic

lightings used in the medieval era. For also improving

the rendering of the complete sculpture, the canopy

and all the gilded ornaments will be added to the 3D

model along with photon mapping and texturing. The

second part of that project will gather all the acquired

new results and will also produce a new video movie

translated in many languages.

ACKNOWLEDGEMENTS

In this pluridisciplinary project the authors were

greatly helped by the following colleagues, in Ecole

Centrale Paris by Franc¸ois-Xavier de Contencin for

3D digitization, Anna Zymla for material analy-

sis, Philippe Denizet and Marie-France Monanges

for video assisted by many students. Colleagues

in Laboratoire de Recherche des Monuments His-

toriques, Vincent Detalle, Olivier Rolland and Annick

VIRTUAL RESTORATION OF A MEDIEVAL POLYCHROME SCULPTURE - Experimentation, Modelization,

Validation and Visualization in Spectral Ray-tracing

469

Texier worked on microstratigraphy, paint analysis

and gilding. The Centre des Monuments Nationaux

and the Saint-Denis basilica scientiﬁc and history

team with Jacqueline Maill´e, Georges Puchal, Robert

Lequ´ement, Alain Erlande Brandenbourg, Franc¸oise

Perrot and Serge Santos. In Louvre museum, Pierre-

Yves Le Pogam gave access to an essential and origi-

nal colored piece while Nicolas Hueber at Ecole Na-

tionale Sup´erieure d’Arts et M´etiers realized the phys-

ical replica assisted by the sculptor Brigitte Bonnet.

Gilles Rafﬁer from AXIATEC facilitated the physical

realization and sponsorship.

REFERENCES

Baxter, W., Wendt, J., and Lin, M. C. (2004). Im-

pasto: A realistic, interactive model for paint. In In

Proc. NPAR2004,the 3rd Intern. Symposium on Non-

photorealistic Animation and Rendering, pages 45–

56.

Born, M. and Wolf, E. (1975). Principles of Optics–

Electromagnetic Theory of Propagation, Interference

and Diffraction of Light–. Pergamon Press, Oxford.

Callet, P. (2006). Metals, alloys and lighting: optical prop-

erties and spectral simulation in computer graphics.

In CIE, editor, Proceedings of the CIE Expert Sym-

posium on Visual Appearance, CIE x032:2007. CIE.

Held in Mus´eum National d’Histoire Naturelle, octo-

ber 19-20th, Paris.

Callet, P. (2007). Couleur et apparence visuelle ii : l’aspect

m´etallique. In Sciences fondamentales, number

AF3253 in AF, page 23. Techniques de l’Ing´enieur.

Callet, P. (mars 1998). Couleur-lumi`ere, couleur-mati`ere

– Interaction lumi`ere-mati`ere et synth`ese d’images.

Collection Sciences en actes, ISBN : 2-84-352-087-

8. Diderot Editeur, Arts et Sciences, Paris, New-York.

CDROM d’images inclus, 320p.

Callet, P. and Zymla, A. (2004). Rendering of binary al-

loys - example and validation with bronze. In team, I.,

editor, Proceedings of ICCVG 04. ICCVG.

de Guilhermy, B. F. (1848). Monographie de l’´eglise royale

de Saint-Denis. Tombeaux et ﬁgures historiques,

Paris, p.240-244. ” ”.

Devlin, K., Chalmers, A., Wilkie, A., and Purgathofer, W.

(2002). Tone reproduction and physically based spec-

tral rendering. In EUROGRAPHICS 2002, STAR -

State of The Art Report.

Dumazet, S., Ariane Genty, a. A. Zymla., de Contencin, F.-

X., Texier, A., Ruscassier, N., Bonnet, B., and Cal-

let, P. (2007). Inﬂuence of the substrate colour on

the visual appearance of gilded sculptures. In Ogleby,

C., editor, Proceedings of the XXIth CIPA conference,

Athens. ISPRS.

Erlande-Brandenburg, A. (1975). Le Roi est mort, Etude

sur les fun´erailles, les s´epultures et les tombeaux des

rois de France jusqu‘`a la ﬁn du XIIIeme si`ecle, Paris.

Biblioth`eque de la Soci´et´e franc¸aise d’arch´eologie,7.

Escalopier, C. (2004). Essai sur divers arts. Th´eophilus

moine artisan du XIIe si`ecle. Texte int´egral traduit du

latin par Charles Escalopier. Paleo.

Gondek, J. S., Meyer, G. W., and Newman, J. G. (1994).

Wavelength dependent reﬂectance functions. In Pro-

ceedings of SIGGRAPH’94, pages 213–220.

Haase, C. S. and Meyer, G. W. (1992). Modeling pigmented

materials for realistic image synthesis. In ACM Trans-

actions on Graphics, pages 305–335.

Kubelka, P. and Munk, F. (1931). Ein beitrag zur optik

der farbanstriche. Zeitschrift fr Technische Physik,

12:593–601.

Millin, A. L. (1791). Antiquit´es nationales, ou Recueil

de monuments pour servir `a l’histoire, tome 2, Paris,

p.10-12,pl.4,ﬁg.2 et pl.5, ﬁg.1. Antiquit´es nationales.

Pitzalis, D., Lahanier, C., Aitken, G., Pillay, R., rodriguez

Echavarria, K., and Arnold, D. B. (2007). 3d tech-

niques to create interactive virtual museums: the state

of the art in the epoch noe. In Jos´e Braz, P.-P. V. and

ao Madeiras Pereira, J., editors, GRAPP 2007 – Sec-

ond Conference on Computer Graphics Theory and

Applications, pages 197–203. INSTICC Press.

Rougeron, G. and Peroche, B. (1997). An adaptive rep-

resentation of spectral data for reﬂectance computa-

tions. In Proceedings of the 8th Eurographics.

Sun, Y., Fracchia, F. D., and Drew, M. S. (2000). A com-

posite spectral model and its applications, scottsdale,

arizona. In Eighth Color Imaging Conference, pages

102–107.

Volz, H. G. and Teague, B. (2001). Industrial Color Testing:

Fundamentals and Techniques (2nd edition) ISBN:

3527304363. Vch Verlagsgesellschaft Mbh.

GRAPP 2008 - International Conference on Computer Graphics Theory and Applications

470