Shows that Hockney's projection method does not significantly aid in the rendering of color and shading in paintings, key properties of the "opticality" that motivated the theory; shows that direct solar illumination is required, which is implausible for many works in the debate. (pdf)[BibTex]

Rebuts Hockney and Falco's suggestion that the concave mirror in the Arnolfini portrait could be turned around and used for projection in the Arnolfini portrait; rebuts their claim that the chandelier "is in perfect perspective" and thus consistent with a projection; gives a more plausible non-optical explanation of the "break" in the trellis in the right-hand panel of the Mérode Altarpiece. (pdf)[BibTex]

Presents several alternate non-optical explanations for the visual features in the Albergati portrait, and points to physical evidence supporting mechanical (not optical) copying and enlarging by compasso da reduzione or Reductionszirkel, evidence that plays no role whatsoever in the optical projection explanation. (pdf)[BibTex]

A popular presentation of the rebuttals to Hockney's theory concerning van Eyck's Arnolfini portrait. (pdf)[BibTex]

Shows that the perspective evidence in Flower still-life supports the claim that both the front and the back of the carpet are not in proper perspective is more salient than the evidence that is, at best, consistent with the use of optics, thereby rebutting Hockney and Falco's claim this painting was created using projections. (pdf)[BibTex]

A rigorous demonstration that the perspective in the Arnolfini chandelier is not "in perfect perspective," thereby undercutting Hockney's claim it was painted using projections; shows that the physical chandelier would have to have been implausibly deformed compared to surviving chandeliers to be even consistent with the projection claims; shows that talented realist artists can paint a complex chandelier "by eye" in better perspective than the Arnolfini chandelier. (pdf)[BibTex]

Uses cast-shadow analysis to rebut Hockney's claim that de la Tour used bright illumination from "outside the frame" or "in place of the other figure" and optical projections when painting his "nocturne" works. (pdf)[BibTex]

An overview of the case for and against Hockney's theory.(pdf)[BibTex]

Shows that the key assumption of Hockney and Falco—that the physical carpet was symmetric—is likely false; explains perspective anomalies as precisely as the optical theory but without requiring complex and undocumented optics. (pdf)[BibTex]

Describes the asymmetry in "Lotto carpets" surviving in museum collections (rejecting Hockney and Falco's key assumption that it was symmetric); provides several non-optical explanations for the "blur" in the painting. (pdf)[BibTex]

Describes pinprick holes discovered in the Albergati silverpoint consistent with the use of a mechanical device such as a Reductionszirkel, but not with the use of optics; describes how the visual evidence in the works is more consistent with mechanical rather than optical copying/enlarging; describes experimental "re-enactment" of the copying by Reductionszirkel. (pdf)[BibTex]

A popular presentation of rebuttals to Hockney and Falco on two key works of Jan van Eyck.(pdf)[BibTex]

An abridged German translation of the Scientific American article. (pdf)[BibTex]

An abridged French translation of the Scientific American article. (pdf)[BibTex]

A review of the proceedings of a four-day workshop devoted to exploring Hockney's theory. (pdf)[BibTex]

The first application of the occluding-contour algorithm to the problem of inferring the direction of illumination in realist paintings. Reconfirms the shadow analysis and scholarly consensus that the illumination in Georges de la Tour's Christ in the carpenter's studio is the candle, rebutting Hockney's claim that the painting was executed using an optical projector. (pdf)[BibTex]

The Bayesian statistical method for integrating various classes of information about the illumination in images including realist paintings such as the works of Caravaggio and Georges de la Tour. (pdf)[BibTex]

An invited tutorial on rigorous analysis of perspective in paintings; shows perspective of a painting of a chandelier done entirely by eye is superior to that of the Arnolfini chandelier, thereby rebutting the claim of Hockney and Falco that the Arnolfini chandelier was painted using optical projections. (pdf)[BibTex]

An invited tutorial on finding the illuminant in realist paintings, such as works of Georges de la Tour. Shows that the image evidence overwhelming refutes Hockney's unorthodox claim that Christ in the carpenter's studio was painted with the light source in place of the figures rather than, say, the location of the candle. (pdf)[BibTex]

An invited tutorial on Chamfer distance and other methods of quantifying form, as in drawings, with application to Jan van Eyck's Albergati portrait. (pdf)[BibTex]

Shows that a talented realist artist using mechanical (not optical) methods from the 15th century can copy Jan van Eyck's Albergati portrait with a fidelity roughly the same as van Eyck. (pdf)[BibTex]

Shows that the constraints upon the putative optical projector used in Lorenzo Lotto's Husband and wife severely undercut the Hockney/Falco explanation for the use of optical projections. (pdf)[BibTex]

A short announcement for the inaugural symposium, Computer image analysis in the study of art. (pdf)[BibTex]

Describes the "De-picting algorithm," which digitally removes layers of brushstrokes to reveal the stages of development of a painting. (pdf)[BibTex]

Uses computer graphics to construct a tableau virtuel of the diptych to reveal that the convex mirror was likely added as an afterthought. (pdf)[BibTex]

Shows that the presence of a 116-cm-wide canvas (screen) in Lotto's putative projector precludes the kind of depth-of-field problems central to the Hockney/Falco optical explanation for this work, thereby rebutting their claim.(pdf)[BibTex]

Shows the high agreement between direction of illumination estimated from cast shadows, occluding contours, shape-from-shading, and computer graphics methods on this Vermeer portrait. (pdf)[BibTex]

Uses computer graphics to show that the illumination in this painting was at the candle and not, as Hockney claims, "outside the frame" or "in place of the other figure." Shows that the cast shadow of Christ's left shin is far more consistent with the illuminant in place of the candle than with Hockney's claim that it was "in place of the other figure [St. Joseph]," thereby corroborating previous analyses rejecting the optical tracing claim for this painting. (pdf)

The first symposium proceedings in this young field. (Order here.)

An overview of computer image analysis for art scholars, focusing on the range of problems addressed, rather than the technical details of how the algorithms work. (pdf)[BibTex]

A weakly model-dependent method for locating an illuminant based on the pattern of lightness over planar surfaces such as a floor or wall. The method shows that the illumination in Georges de la Tour's Christ in the carpenter's studio and Caravaggio's The calling of St. Matthew is local, and not distant solar illumination, thereby rebutting Hockney's claim that these paintings were executed by means of optical projections. (pdf)[BibTex]

An expert peer-reviewed journal article showing that professional artists using mechanical aids (such as a Reductionzirkel) can achieve fidelity superior to that found in van Eyck's oil work, and that such mechanical methods can also explain the displaced and scaled ear. These results show that van Eyck did not need to use the most complicated optical system of its day (had it even existed) but that documented methods from that time can explain all the relevant visual evidence. This, and the Ketelsen et al discovery of pinprick holes in the silverpoint (which have no role whatsoever in the optical explaination), argue overwhelmingly for a mechanical (not optical) explaination of van Eyck's copying. (pdf)[BibTex]

Extends the occluding-contour algorithm to the case of diffuse lighting. This more powerful algorithm exposes that different portions of a realist painting were executed under different illumination conditions, the analog of "compositing" in a digital photograph (pdf)[BibTex].

A short note on the restoration and computer search for repetitive patterns in the brick work to better understand the artist's working methods. (pdf)[BibTex]

Computer graphics renderings of Caravaggio's The calling of St. Matthew reveal subtleties in the lighting. (pdf forthcoming)

We infer the position of illuminants in paintings by Georges de la Tour and Caravaggio based on the pattern of luminance over the floor and wall, all to test whether these artists employed optical projections during the execution of their works. (pdf)

A full computer correlation analysis of the brickwork in this painting to test whether the artist used 'counterproofing,' that is, pressed a printed image onto the painting. (pdf)

We reconstruct the three-dimensional tableau in Fraser's painting based on the multiple views depicted in the mirrors in the painting. (pdf)

We explore standard machine learning techniques to the authentication of Jackson Pollock drip paintings based on a number of visual features beyond the traditional "fractal" features. (pdf)

An edited volume on computer vision applied to art (order here).

A computer graphics reconstruction of the tableau in the painting as well as the viewer's space outside the painting gives a new understanding of sightlines and perspective. (pdf)

The use of correlation analysis to search for repeated brick patterns which might indicate the use of counterproofing. (poster pdf, forthcoming)

xxxxxxx (pdf)

Content

Techniques for inferring the "camera" parameters (viewing positions) from both correspondences in reflected images depicted within a realist painting and from the geometry of mirrors. (pdf)[BibTex]

An overview of the key publications in computer vision applied to the study of art. (pdf)[BibTex]

The proceedings of a conference on computer vision applied to the study of art. (purchase here)

Content

A summary of the unanimous scholarly rejection of the Hockney/Falco theory for the early Renaissance. (pdf forthcoming)

An overview of the methods and applications of computer graphics in the study of realist fine art. (pdf forthcoming)

A presention on how computer methods can reveal perceptual inaccuracies in perspective and lighting, and how computer graphics can reveal artists' studios, a new domain for perceptual studies in the arts. (pdf)

A description of the use of image feature extraction and sophisticated machine learning methods for inferring the compositional principles of Piet Mondrian's neo-plastic abstract paintings, including generating pseudo-Mondrians. (pdf forthcoming)

A description of the use of multiple visual features and methods of feature pruning from pattern classification to the problem of authentication of Jackson Pollock's drip paintings. (pdf)

A description of the use of scale and orientation insensitive correlation analysis between paintings by Jan van der Heyden in search of repeated brick patterns, indications of counterproofing. (pdf)

A description of the use of sophisticated bi-directional reflectance functions (BRDFs) for inferring the lighting in Caravaggio's studio. (pdf forthcoming)

A rigorous investigation into the central claim in support of Jan Vermeer's putative use of the camera obscura for tracing projected images. (pdf forthcoming)

A computer graphics reconstruction of the studio in Parmigianino's work, exposing the artistic liberties and expressive use of distortion. (pdf forthcoming)