Conference PaperPDF Available

Beurs' historical recipe and material perception of grapes in Dutch Golden Age still-lifes

Authors:

Abstract and Figures

Intensive observation of the world, and the intention of realistically transferring it to the canvas, allowed Dutch Golden Age painters to develop an implicit knowledge of the visual patterns people use to infer different materials, imitating key optical phenomena via shortcuts. To understand the origin of the astonishing realism of Dutch 17 th century paintings, we refer to the treatise of Willem Beurs, "The Big World Painted Small", a precious source of technical information about oil painting. One of the questions we aim to answer is: how did they produce such true-to-life depictions? We chose the representation of grapes as case study, due to the simultaneous presence and interaction of different material properties, like glossiness, translucency and bloom. Glossiness and translucency are of primary importance in vision science. Thus, understanding their rendering and perception for the case of grapes, can lay the groundwork for a more general theory of gloss and translucency. We investigated if the material properties proposed by Beurs to paint grapes are actually perceived in paintings, and how they relate to their perceived convincingness. Among these material qualities, we took a closer look at glossiness and tried to predict its perception via image statistics of specular reflections.
Content may be subject to copyright.
Beurs’ historical recipe and material perception of grapes in
Dutch Golden Age still-lifes
Francesca Di Cicco, Maarten Wijntjes, Sylvia Pont; Perceptual Intelligence Lab, Faculty of Industrial Design Engineering; Delft
University of Technology; Delft, The Netherlands
Abstract
Intensive observation of the world, and the intention of
realistically transferring it to the canvas, allowed Dutch Golden
Age painters to develop an implicit knowledge of the visual
patterns people use to infer different materials, imitating key
optical phenomena via shortcuts. To understand the origin of the
astonishing realism of Dutch 17th century paintings, we refer to the
treatise of Willem Beurs, “The Big World Painted Small”, a
precious source of technical information about oil painting. One of
the questions we aim to answer is: how did they produce such true-
to-life depictions?
We chose the representation of grapes as case study, due to
the simultaneous presence and interaction of different material
properties, like glossiness, translucency and bloom. Glossiness and
translucency are of primary importance in vision science. Thus,
understanding their rendering and perception for the case of
grapes, can lay the groundwork for a more general theory of gloss
and translucency.
We investigated if the material properties proposed by Beurs
to paint grapes are actually perceived in paintings, and how they
relate to their perceived convincingness. Among these material
qualities, we took a closer look at glossiness and tried to predict its
perception via image statistics of specular reflections.
Introduction
The research presented in this paper is part of the
interdisciplinary NICAS project “Recipes and Realities”. In this
project, we aim to unveil the secret behind the marvelous rendering
of materials, typical of Dutch Golden Age (still-life) paintings. We
approach the problem from two sides: one is technical art history,
i.e. the scientific analysis of artworks with the aid of written
historical sources, and the other is visual perception.
The investigation of historical records can facilitate the
understanding of the process of art making. In the case of Dutch
Golden Age paintings, a precious source of technical information
is the collection of recipes for oil painting written by Willem Beurs
in 1692 1, 2. His treatise, “The big world painted small” (Figure 1),
allows to access the pictorial practice of the time 3. This treatise
consists of six chapters, containing instructions on how to render
all sorts of material and surface effects. It is the first written source
completely devoted to oil painting, to the materials, their
preparation and their use. Willem Beurs (1656- after 1692) was a
painter who worked, throughout his life, on the major categories of
painting: landscape, portrait and still life 3. He owned, therefore,
the knowledge and the experience for teaching painting, and his
book had indeed educational purposes.
Figure 1. Page of the title of Willem Beurs treatise:The big world painted
small” (1692) 1.
In order to learn to replicate the world, nature is the best
teacher, as Beurs stressed often. Observation is certainly pivotal to
reproduce on the canvas what surrounds us in the most realistic
way. However, observation, or painting after life (naer het leven) 4,
was not the only method used in the 17th century.
Standard compositions and patterns, as the ones taught by
Beurs himself, which return in the painting practice of the Old
Masters, reveal the imitative process of painting “from the mind”
(uyt den gheest) 4. A famous example of such artificial
compositions, is the Vase of Flowers in a Window by Bosschaert
(Figure 2). The flowers are all meticulously and exactly
reproduced, but they belong to different seasons. It was thus
impossible that Bosschaert painted the bouquet from life.
Moreover, as remarked by Westermann 4, the mountains visible in
the background definitely do not represent a real Dutch landscape.
Another hint that this painting is the outcome of an imaginary
composition, lies in the importance given to each single flower.
Slive 5 pointed out how each flower appears as a single portrait,
being all of them lit in the same way, none is left in the shadow
and they are all main characters of the scene.
IS&T International Symposium on Electronic Imaging 2018
Human Vision and Electronic Imaging 2018 536-1
https://doi.org/10.2352/ISSN.2470-1173.2018.14.HVEI-536
© 2018, Society for Imaging Science and Technology
Figure 2. Vase of Flowers in a Window, Ambrosius Bosschaert the Elder
(1618), oil on panel. Downloaded from the online repository of the Mauritshuis,
The Hague.
This is a clear-cut example of the ‘cheating’ technique used
by painters, what Cavanagh called “alternative physics” 6. The true
physics of light is sacrificed by Bosschaert for the sake of the
message, the representation, and most probably to show off his
masterful skills of rendering. Yet, this ‘wrong’ lighting is not
disturbing and does not impair the striking realism of the scene.
Uncover the schemes and conventions that 17th century Dutch
painters adopted to imitate materials, surfaces and textures, is the
final goal of our project.
The source of realism
Painters and scientists from the 17th century shared the eager
desire to better understand the world, leading to important
discoveries in several fields, including optics. ‘Optics’ is intended
here as the interaction of light with different surfaces and its visual
effects, rather than a discussion on the theory of perspective 7, 8, or
on the use of optical devices, such as the camera obscura 9, 10.
The study of light and the ‘art of reflection’, or reflexy-const,
as it was first defined by Karel van Mander in his treatise Het
Schilderboeck (1604) 11, is regarded as the primary source of
realism in 17th century paintings 7, 12. The value of the novel
understanding of optics is acknowledged by Beurs himself, who,
throughout the book, mentions the work of scientists like Huygens,
Descartes and Boyle. However, he also points out that owning such
knowledge of the physical behavior of light, has more an
intellectual value for the painter than a practical one (“as long as
they have a good understanding of the pigments and how to paint
with them”) 2. Indeed, the adoption of oil painting and the
exploitation of its versatile properties may have had the most
important influence on the life-like imitation of nature 13, 14. The
manifold physical-chemical properties of oil, make it the “most
powerful and stable medium”, as noticed by Beurs 2. Among the
most relevant characteristics of the oil medium, there are: its
transparency, the blending possibilities offered by the slowness of
its drying, and the many ranges of fluidity and handling that permit
to create different texture effects.
But not all the oils produce the same visual effect. For
instance, in the recipe to paint bloom on grapes, given by Beurs, it
is stressed to use a “white oil”, probably referring to poppy seed oil
15, since, compared to linseed oil, it becomes yellow to a lesser
extent, ensuring the lasting of the original colors.
Other visual aspects depending on the choice and use of oil
are, for example, the colors’ saturation, the glaze or even the
rendering of transparent and translucent objects 16.
It is the transparency of the oil that allows for the glazing
technique. This technique consists in superimposing one or more
thin transparent layers of colored oil, on top of a dry, opaque
underlayer. The light can penetrate the translucent glaze layer,
bounce off the underlayer, and reach the eye of the observer
carrying modulated colors. The ‘trick’ behind the glazing, is the
optical mixing, filtering and multiple scattering of colors. Via these
mechanisms, the glaze layers are reported to change the hue of the
colors, add visual depthand make the paintings glow16, 17. The
exact physical and perceptual effects of glazing, however, still
have to be uncovered.
Alongside, the optical and chemical properties of pigments
are equally important to the final visual impression. To mention
one crucial property to take into account, the difference of the
refractive index of a pigment, compared to that of the mixing oil,
determines the consequent opacity or transparency of the paint
layer.
It is thus evident that for the most comprehensive perceptual
analysis of a painting, we cannot consider only what is depicted,
but also how it was depicted, relying on the effective combination
of scientific analytical methods, historical sources and
psychophysical experiments.
Literature review
Artworks are goldmines to disclose knowledge about visual
perception. The field of the ‘psychology of art’ was initiated by the
influential writings of Arnheim 18 and Gombrich 19. They both
addressed the problem of the relationship between art and reality,
but using divergent approaches. Arnheim 18 based his reasoning on
the Gestalt theory of perceptual organization, by listing the visual
categories that are arranged as a whole in visual perception and art
production. To Gombrich 19, on the other hand, visual perception
and interpretation is driven by prior knowledge and expectations.
He proposed that such learned experience (“schemata”) is also at
the bottom of artistic creation, which arises from known
conventions and it is then adjusted to match the real world, via
direct observation.
Subsequent works of psychologists and neuroscientists have
explored the learning possibilities offered by art. Zeki 20 and
Cavanagh 6 have referred to artists as ‘neuroscientists’, as being
able to extract the “true character of objects” 20, and render it via
shortcuts that simplify the physics of the world 6.
To Cavanagh 6, images that deviate from the laws of physics
without compromising the convincingness of the representation to
the eye of the observer, are the key to understand the functioning
536-2 IS&T International Symposium on Electronic Imaging 2018
Human Vision and Electronic Imaging 2018
of our visual system. He claimed that unnoticed physical errors in
paintings reveal that the brain makes actually use of a simplified
physics itself, for the sake of perceiving the world in the most
efficient way. Moreover, the conventions of such ‘alternative
physics’ are well established in the human brain since prehistoric
times (like the convention of line drawings of cavemen).
The theory of visual shortcuts exploited by painters is
consistent with the ‘statistical appearance model’, proposed by
Fleming 21 as a third alternative to the two main theories leading
material perception. To Fleming 21 visual perception of materials
should not be treated neither as an ‘inverse optics’ problem 22, nor
as an image statistics question 23, 24. We rather tend to infer the
‘look’ of the materials form their key parameters visible in the
image, and estimate how changing the “appearance attributes”
would change the image 21, 25, 26.
An example of image features diagnostic for material
properties, can be found in the study of Sayim and Cavanagh 27.
They investigated the cues used by artists throughout the centuries
to depict transparency. Apart from the often-used luminance
constraints, that match well the X-junctions theory of Metelli 28,
they also identified material constraints. In particular, the material
property of glossiness can be diagnostic for transparency. They
showed 27 that, in the pictorial practice, placing highlights on the
surface of a highly transparent material, can constitute the most
revealing cue of its transparency.
The case study of grapes
In Beurs’ treatise, one of the six chapters is devoted to still-
life, and it opens with the recipe for rendering grapes, “pleasing to
the eye and a treat for the tongue, and containing the juice that,
when used well, gives joy to God and humankind2. But grapes
are not only the sacred fruit of Bacchus, symbol of abundance and
fertility. According to Roger de Piles, the bunch of grapes
constitutes the metaphor of a painting. By observing a bunch of
grapes, one can learn the best distribution of light and shadows to
render chiaroscuro, and the sense of unity of the composition 29, 30.
This theory is exemplified in Figure 3.
Figure 3. Illustration from “The principles of painting” (Roger de Piles, edition
of the 1743).
From a visual perception point of view, the case of grapes is
particularly interesting due to the complex combination of different
material properties. From daily experience with real bunches of
grapes, we know that grapes are translucent and glossy, but can
also be (partly) covered by a matte layer of bloom (a whitish waxy
layer on the surface of the grapes). Thus, modeling an optical
function to convincingly render grapes can be a computational
nightmare.
The knowledge of which material properties are necessary to
render grapes, and how they should be combined, is again
something we can learn by reading Beurs. In his recipe, he
provides explicit instructions about which pigment needs to be
applied to paint each part of the bunch. These instructions point
also out, more or less implicitly, the different material properties
and their cues. For instance, he advices to “give a sheen on the
midtone with white gently blended in” 2. In these few words Beurs
is telling us to render the property of glossiness by applying a
highlight on the part of the surface where no bloom was painted
(the midtone), thus indicating how glossiness and bloom should be
combined. Moreover, he suggests that the cue for glossiness, i.e.
the highlight, should be white and not too sharp.
The examination of historical recipes is an incomparable
source of information, not only to shed light on the studio practice
13, 31, but also to analyse and reconstruct the artworks (e.g. Stols-
Witlox studied the recipes of grounds and preparatory layers,
gleaning also from Beurs) 32. Differences between artists, in the
process of making and in the choice of pigments, can result in
different ways of rendering materials and in a different perceptual
experience. Wallert 15 has examined the cross sections of several
Dutch 17th century masterpieces and compared them with the
instructions given by Beurs.
For what concerns grapes, being them either white, blue or
red, Wallert 15 found that a common practice to represent the
bloom was a mix of ultramarine, lead white and lake, as also
described by Beurs. Such recipe was found in Festoon of fruits and
flowers by Jan Davidsz. de Heem (Figure 4), Still life with a golden
goblet by Pieter de Ring (Figure 5), and in Still life with fruit,
oysters and a porcelain bowl by Abraham Mignon (Figure 6). De
Heem and Mignon, also share what Wallert calls “a systematic
way” to build the grapes, starting with the lit side (‘the day’) and
the shadows, continuing with the bloom and refined with the
highlights, again matching what Beurs described. Interestingly, this
was not the procedure followed, for example, by Coenraet Roepel
in his Still life with fruit (Figure 7). His white grapes were painted
by applying a thin layer of green paint on the ground, made of lead
white, Prussian blue and yellow lake. He then added the white
highlights and finished with the bloom.
This difference in the process of making and in the choice of
pigments, results in a different way of rendering the grapes and in
different perceptual outcomes.
IS&T International Symposium on Electronic Imaging 2018
Human Vision and Electronic Imaging 2018 536-3
Figure 4. Festoon of fruit and flowers, Jan Davidsz. de Heem (1660-1670), oil
on canvas. Downloaded from the online repository of the Rijksmuseum,
Amsterdam.
Figure 5. Still life with Golden Goblet, Pieter de Ring (1640-1660), oil on
canvas. Downloaded from the online repository of the Rijksmuseum,
Amsterdam.
Figure 6. Still life with Fruit, Oysters and a Porcelain Bowl, Abraham Mignon
(1660-1679), oil on canvas, Downloaded from the online repository of the
Rijksmuseum, Amsterdam.
Figure 7. Still life with Fruit, Coenraet Roepel (1721), oil on canvas.
Downloaded from the online repository of the Rijksmuseum, Amsterdam.
536-4 IS&T International Symposium on Electronic Imaging 2018
Human Vision and Electronic Imaging 2018
Via psychophysical rating experiments, we first investigated
if the material properties proposed by Beurs (i.e. translucency,
glossiness, bloom and three dimensionality), were actually
perceived in the grapes of several 17th century paintings. We also
tested how each of these properties relates to the final
convincingness. We did not find a prototypical grape appearance
but instead a wide range of ratings. However, we found that all the
properties prescribed by Beurs were necessary ingredients for the
overall convincingness, even though the weight of their
contribution was case-dependent. In Figure 8, we illustrate this
finding by deleting one or another of the attributes from a bunch of
grapes that was generally perceived as highly convincing. The top
image (A) shows a detail of the original painting, exhibiting each
of the four properties. The two images below were modified using
Photoshop (CC 2017.0.1). From one (B) we deleted the highlights,
thus eliminating the glossiness, whereas in the other image (C),
bloom was deleted. In both cases there is a noticeable drop in
convincingness.
Once we determined that all the material properties listed in
Beurs’ recipe contribute to enhance the realistic representation of
grapes, we looked into the shortcuts that reveal gloss perception.
We developed a novel method to compute the low-level cues of
highlights (contrast, coverage and sharpness), reported to be
diagnostic for glossiness 25. Via segmentation analysis, we could
perform the cues’ computation directly from the images.
A)
B)
C)
Figure 8. Detail of Still Life with Fruit and Oysters, Abraham Mignon (1660-
1679), oil on canvas. Downloaded from the online repository of the
Rijksmuseum, Amsterdam. A) is the original version; B) and C) are the
modified versions (B: no highlights; C: no bloom).
Acknowledgments
This work is part of the research program NICAS “Recipes
and Realities” with project number 628.007.005, which is partly
financed by the Netherlands Organization for Scientific Research
(NWO) and partly by Delft University of Technology. Maarten
Wijntjes was financed by the VIDI project “Visual communication
of material properties”, number 276.54.001.
References
[1] W. Beurs, De groote waereld in 't kleen geschildert, of schilderagtig
tafereel van 's Weerelds schilderyen. Kortelijk vervat in ses boeken.
IS&T International Symposium on Electronic Imaging 2018
Human Vision and Electronic Imaging 2018 536-5
Verklarende de hooftverwen, haare verscheide mengelingen in oly en
der zelver gebruik. Amsterdam, van Waesberge, 1692.
[2] W. Beurs, The big world painted small, translated by Myra Scholz
and with introductory essays by Jeroen Stumpel, Ann-Sophie
Lehmann, Lisa Wiersma and Erma Hermens, The Getty Research
Institute, expected publication 2018.
[3] A. Wallert, “De Groote Waereld in ‘t Kleen Geschildert (The Big
World Painted Small): a Dutch 17th century treatise on oil painting
technique, in Eyb-Green S., Townsend J., 2012.
[4] M. Westermann, A Worldly Art. The Dutch Republic, 1585-1718,
New Haven: Yale University Press, 2005.
[5] S. Slive, Dutch Painting, 1600-1800, New Haven and London: Yale
Press University, 1998.
[6] P. Cavanagh, The artist as neuroscientist”, Nature, 434, 301-307,
2005.
[7] S. Dupré, "The historiography of perspective and reflexy-const in
Netherlandish art”, Nederlands Kunsthistorisch Jaarboek, 61, 35-60,
2011.
[8] H. Ono, N.J. Wade, & L. Lillakas, “The pursuit of Leonardo’s
constraint”, Perception, 31, 83-102, 2002.
[9] D.G. Stork, Did early Renaissance masters use optical projections
while painting?”, in Frontiers in Optics 2004/Laser Science
XXII/Diffractive Optics and Micro-Optics/Optical Fabrication and
Testing, OSA Technical Digest Series (Optical Society of America,
2004), paper FWW5, 2004.
[10] D.G. Stork, “Did Hans Memling employ optical projections when
painting "Flower Still-Life?"”, Leonardo, 38(2), 155-160, 2005.
[11] W.S. Melion, Shaping the Netherlandish Canon: Karel Van Mander's
Schilder-Boeck, University of Chicago Press, 1991.
[12] L. Pincus, Painting Light. Artifice and reflexy-const in the Dutch
seventeenth century”, Hollands Licht, 192, 141-150, 2011.
[13] A.-S. Lehmann, Fleshing out the body, The ‘colours of the naked’ in
workshop practice and art theory, 1400 1600”, Body and
Embodiment in Netherlandish Art, Netherlands Yearbook for History
of Art 58, ed. By A. Lehmann & H. Roodenburg, (Zwolle: Waanders
2008).
[14] A.L.S. Vandivere, “From the ground up. Surface and sub-surface
effects in fifteenth- and sixteenth-century Netherlandish paintings”
(PhD thesis), Universiteit van Amsterdam, Amsterdam, The
Netherlands, 2013.
[15] A. Wallert, Still lifes: techniques and style. An examination of
paintings from the Rijksmusuem, Amsterdam, Rijksmuseum,
Waanders, 1999.
[16] A.-S. Lehmann, “The matter of the medium: some tools for an art-
theoretical interpretation of materials”, in The Matter of Art:
Materials, Technologies, Meanings 1200-1700, ed. by C. Anderson,
A. Dunlop, P.H. Smith (eds.), Manchester: Manchester University
Press, 2015.
[17] E. van Duijn, & J. Roeders, Gold-brocaded velvets in paintings by
Cornelis Engebrechtsz.”, Journal of Historians of Netherlandish Art,
4(1), 1-28, 2012.
[18] R. Arnheim, Art and Visual Percpetion: A Psychology of the Creative
Eye, Berkley, CA: University of California Press, 1954.
[19] E. Gombrich, Art and Illusion: A Study in the Psychology of Pictorial
Representation, London: Phaidon Press, 1960.
[20] S. Zeki, “Art and the brain”, Daedalus, 127(2), 71-103, 1998.
[21] R.W. Fleming, “Visual perception of materials and their properties”,
Vision Research, 94, 62-75, 2014.
[22] Z. Pizlo, “Perception viewed as an inverse problem”, Vision
Research, 41(24), 3145-3161, 2001.
[23] I. Motoyoshi, S. Nishida, L. Sharan, & E. H. Adelson, “Image
statistics and the perception of surface qualities”, Nature, 447(7141),
206-209, 2007.
[24] S. Nishida, & M. Shinya, “Use of image-based information in
judgments of surface reflectance properties”, Journal of the Optical
Society of America A, 15, 2951-2965, 1998.
[25] P. J. Marlow, J. Kim, & B. L. Anderson, The perception and
misperception of specular reflectance”, Current Biology, 22, 1909
1913, 2012.
[26] P. J. Marlow, & B. L. Anderson, “Generative constraints on image
cues for perceived gloss”, Journal of Vision, 13(14), 1-23, 2013.
[27] B. Sayim, & P. Cavanagh, “The art of transparency”, i-Perception, 2,
679-696, 2011.
[28] F. Metelli, “The perception of transparency”, Scientific American,
230, 90-98, 1974.
[29] G.W.R. Ward, The grove encyclopedia of materials and techniques in
art, New York: Oxford University Press, 2008.
[30] M. Barasch, Theories of art: 1. From Plato to Winckelmann, New
York: Routledge, 2000.
[31] P. H. Smith, & T. Beentjes, Nature and art, making and knowing:
reconstructing sixteenth-century life-casting techniques”,
Renaissance Quarterly, 63(1), 128-179, 2010.
[32] M. Stols-Witlox, ‘From reading to painting’: authors and audiences
of Dutch recipes for preparatory layers for oil painting”, Early
Modern Low Countries, 1, 71-134, 2017.
536-6 IS&T International Symposium on Electronic Imaging 2018
Human Vision and Electronic Imaging 2018
Thesis
Full-text available
This thesis explores convincing stuff depicted in 17th century paintings, with the primary aim of understanding their visual perception. ”Stuff” is the term first introduced by Edward Adelson in 2001 to differentiate materials from objects, and to call attention on the research gap in material perception. In an interesting parallel, the representation of materials in paintings constitutes a knowledge gap in art history as well. Both gaps have only recently been recognized and started to be addressed in their respective research fields. In this thesis, representation and perception come together to create a virtuous circle in which the knowledge of painters about the representation of materials is used to understand the mechanisms of the visual system for material perception, and this is in turn used to explain the pictorial features that make the representation of materials so convincing. The common thread used here to connect representation and perception, is ”The big world painted small”, a long-forgotten booklet of pictorial recipes written by the Dutch painter Willem Beurs in 1692. We argue that this book represents an index of key features for material perception, that means an index of image features that always work as perceptual cues regardless of the illumination and the viewing conditions of the depicted scene. The main research objective of this dissertation is: To understand the convincing depiction and perception of materials in 17th century paintings, connecting the image features found in paintings and listed by Beurs to their role as perceptual cues. In order to achieve this objective, we employed a novel, interdisciplinary research approach, merging science of human and computer vision, technical art history, and the historical textual source of Beurs.
Article
Full-text available
Image structure is generated by distinct physical sources that include 3-D surface geometry, surface reflectance, and the light field. A fundamental problem in midlevel vision involves understanding how, and how well, the visual system separates images into their underlying sources. A large body of recent work has shown that the perception of gloss exhibits significant dependencies on both 3-D surface geometry and the light field in which the surface is embedded. We recently proposed that these dependencies arise from the visual system relying on a restricted set of dimensions of specular image structure that are heuristically used to estimate surface gloss (Marlow, Kim, & Anderson, 2012). Here, we report a series of studies that manipulate surface geometry and the structure of the light field to prospectively test whether these image cues predict perceived gloss for a broad range of surface geometries and light fields. We identify generative constraints on each of our proposed cues that allow us to predict which cue will vary most for a given surface geometry, reflectance function, and light field. Our psychophysical data reveal that our generative manipulations succeed in promoting the significance of the intended cue, which in turn increases their predictive value in observers' judgments of gloss. Our results suggest that the perception of gloss in static, monocular images relies on a heuristic weighting of cues to specular image structure, and offer a unified explanation of why surface geometry and the light field modulate perceived gloss.
Article
Full-text available
Misidentifying materials-such as mistaking soap for pâté, or vice versa-could lead to some pretty messy mishaps. Fortunately, we rarely suffer such indignities, thanks largely to our outstanding ability to recognize materials-and identify their properties-by sight. In everyday life, we encounter an enormous variety of materials, which we usually distinguish effortlessly and without error. However, despite its subjective ease, material perception poses the visual system with some unique and significant challenges, because a given material can take on many different appearances depending on the lighting, viewpoint and shape. Here, I use observations from recent research on material perception to outline a general theory of material perception, in which I suggest that the visual system does not actually estimate physical parameters of materials and objects. Instead-I argue-the brain is remarkably adept at building 'statistical generative models' that capture the natural degrees of variation in appearance between samples. For example, when determining perceived glossiness, the brain does not estimate parameters of the BRDF. Instead, it uses a constellation of low- and mid-level image measurements to characterize the extent to which the surface manifests specular reflections. I argue that these 'statistical appearance models' are both more expressive and easier to compute than physical parameters, and therefore represent a powerful middle way between a 'bag of tricks' and 'inverse optics'.