Rosetta Stoned?: Hockney, Falco and the sources of őopticality‚ in

Lorenzo Lotto‚s őHusband and Wife‚


Christopher Tyler

Associate Director

Smith-Kettlewell Eye Research Institute, San Francisco          


In his recent book, David Hockney proposes that the őoptical quality‚ of Flemish art of the early 1400s arose because the artists, van Eyck in particular, suddenly began to use optical devices for the accurate projection onto the canvas.  In presenting this case at a recent symposium on the issue, Hockney described one particular painting as the „Rosetta Stoneš of their argument, because it was the one that allowed the details of the optical hypothesis to be examined most accurately.  That painting was the őHusband and Wife‚ by Lorenzo Lotto (1543).  In it is depicted a tapestry tablecloth with a distinctive octagon feature at the center of the table.  This feature has the curious property that it seems to go out of focus as it recedes from the viewer.  It is argued that this blurring is „proofš that Lotto copied the detail of this pattern from an optical projection of a real tapestry in his studio, validating the idea that optical projection was in widespread use during the Renaissance (as opposed to the well-known eighteenth-century use of the camera obscura by artists such as Canaletto and Joshua Reynolds).



Figure 1.  Lorenzo Lotto  őHusband and Wife‚ (c.1543)


Figure 2.  Detail of the octagonal design on the tapestry


Before considering the plausibility of the specific claim about the Lotto painting, it may be pointed out that the patch of blur on this tapestry makes a weak case for the widespread use of optics, because it is the only Renaissance painting (in southern or northern art) that exhibits this particularity.  Both the geometric perspective of the Italians and the optical and light effects of the Flemish are renowned for the precision of their outlines.  When a freer style came in with Titian, Tintoretto and Rubens, it extended throughout the painting and did not support the concept of a region of sharp focus spreading into blur.  Even if Lotto had used optical projection, this isolated piece of evidence would not support its widespread use.


Rubens    őBathsheba‚   (1556)

These considerations question the concept of an őoptical look‚ that plays such a role in Hockney‚s account.  He associates the idea of an optical look with the high accuracy and strong shadowing of such artists as van Eyck and Caravaggio.  In fact, however, Hockney‚s own demonstrations (with David Graves) both for the book and at the New York conference presenting these ideas, made it clear that the kind of optics available to Renaissance artists would have had a narrow depth of focus and a large degree of blurring in objects slightly outside the focused region.  The look of an optical projection is thus a pronounced fluctuation between sharp and soft focus throughout the painting.  The valid őoptical look‚ would, indeed, have been to paint this fluctuating focus, as is typically seen in the photo-realist painters of our own times.  No such fluctuating focus is seen in any other Renaissance works, and seems to first appear in the 17th-century paintings of Vermeer.


To address the discrepancy, Hockney and his collaborator Charles Falco, an optical specialist at the University of Arizona, propose that the artists who aimed at the optical look frequently changed the position of the lens to refocus on each area of the scene, so as to generate an image that matches our perceptual experience of clear focus in all regions of an observed scene.  The telltale sign of this refocusing is, they suggest, the minor inaccuracies in the geometry of the projection that they discover in several paintings.  Lines that should be continuous in objects and textures seem to show shifts in angle, with reconvergence to slightly different vanishing points.  This reconvergence is a second line of evidence that some Renaissance masters used optical projection in their paintings.  Hockney mentions that he was alerted to this effect by the slight discrepancies between frames in his own photocollage experiments.


One problem with this optical story is that the two types of evidence are mutually contradictory when they are found in the same painting, such as the Lotto őHusband and Wife‚.  If Lotto readjusted an optical projection in order to avoid blurring of one region of the painting, why did he assiduously paint the blur in another (central) region?  Conversely, if he wished to use the blur to enhance the depth impression, why did he not employ it in the receding regions of the tablecloth (where the pattern is actually depicted with high clarity)?  Hockney and Falco do not address these inherent contradictions in the interpretation of their „Rosetta Stoneš.



Figure 4.   Reconstruction of tapestry vanishing points


To press the case, we may follow Falco‚s geometric analysis to its logical conclusion to provide a complete reconstruction of the central octagon in the painting.  Because it recedes into blur, this particular region of the pattern must have been painted without readjustment of its optical projection, on their hypothesis (Hockney & Falco, 2000).  This region of the painting should have been copied to exactly match the optical projection.  It follows that the geometry within this pattern element should be perfectly coherent, exactly adhering to the laws of perspective projection.  Technically, these laws imply that i) each sets of parallel lines with this octagonal pattern should project to a single vanishing point and ii) that all the different the vanishing points should lie at the horizon (assuming that the table is horizontal).  This alignment of vanishing points along the horizon is a geometric rule of perspective that must be followed by any set of undistorted optical projections. 


A plausible horizon is shown for the projection lines along the front of the table (Figure 4, white lines).  As long as the canvas remains in the same position, all horizontal vanishing points for different lens positions should project to the same horizon.  For the vanishing-point horizons to vary, the canvas would either have to be shifted vertically or slanted at different vertical angles.  Neither change seems likely when an artist is attempting to keep the scene registered on the canvas during shifts in the focal distance of a projecting lens.  (The distortion of a pattern away from the center of the lens would, at worst, only add a gradual curve to the horizon and the associated line of vanishing points.) One look at the projections of the various parallels within the octagons shows that the conspicuously fail to adhere to any consistent horizon.


The alternative to the optical projection hypothesis is that Lotto drew the pattern casually, with only sufficient accuracy to give the impression of a deep-pile tapestry as a background for his portrait, and was not concerned with a geometric accuracy that might withstand examination half a millennium into the future.  On this interpretation, the locations of the vanishing points would be haphazard and would not conform to any particular geometric rule.  The reconstruction shown in the detail of Fig. 2 is clearly in support of the non-optical explanation.  Each set of colored lines derive from mutually parallel elements of the pattern and should converge to a single vanishing point, but the convergence for each set is incoherent, with points scattered widely through the painting[i].


Detailed examination reveals another curious feature of the octagonal pattern  Ų  lines connecting symmetrical corners of the pattern are not parallel to each other.  To evaluate this issue clearly, the perspective of the tapestry has been retransformed to shown the pattern in plan view, as if looking down directly at the top of the table (Fig. 5).  The degree of distortion entailed is visible in the anamorphic shape of the man‚s hand at right.  The configuration of octagonal element now becomes clearly visible.  Despite the rectangularity of the pattern around it (white lines), the octagonal motif is severely distorted.  The cyan lines reveal a substantial convergence to the right that remains in upper and lower parts of the motif, although inconsistent with the central section remaining parallel.  The bottom of the őstem‚ of the octagon meets the edging at two discrepant levels (white lines).  None of these geometric misalignments would be possible if the pattern was copied from an optical projection.  It is clear that this octagon is heavily distorted, making it quite unlikely that either optics or perspective geometry were used in the depiction of this tablecloth.

Figure 5.  Flat projection of tapestry with construction lines.


The implication of this geometrical analysis is that Lotto had simply őeyeballed‚ the pattern of the tablecloth, to use Hockney‚s term.  Unlike some of his contemporaries, such as Hans Holbein, Lotto is not an artist who appears to pay particular attention to geometric accuracy.  He rather emphasizes the emotional interplay among the characters depicted.  It makes sense that he would have given only cursory attention to the details of the tablecloth on which his sitters are leaning, approximating the details of the pattern rather than using multiple optical projections.  If Lotto‚s painting is the core of the evidence for the early use of optics, one would have to conclude that the case advanced by Hockney and Falco is weak at best.


When it comes to painting, Hockney does not appear to believe his own proposal that optics were used for slavish copying zone by zone.  In his book, Hockney goes to great lengths to set up a camera obscura with the optical device of a concave mirror and make some sketches of a sitter in Renaissance garb, to demonstrate the feasibility of the optical approach to artistic representation.  The resultant drawings are reproduced in the book.  At the New York conference discussing these issues, however, Hockney was asked why the book did not show any paintings generated by this method.  His reply was that he had tried painting within his camera obscura, but had abandoned the effort „within ten minutesš because it was far too impractical.  He very soon turned to copying the projected image by painting onto a separate canvas rather than trying to mirror the details directly on the plane of projection.  He was sure that any Renaissance artist would have the same experience, turning to the copying approach within ten minutes.


It is obvious that, once he is relying on his eye to match the painted and the projected image, Hockney is in very much the same situation as the traditional artist setting up his easel in the direction of a scene to be copied.  Perhaps the framing and optical qualities would be enhanced by setting up a booth in which the particular region of interest could by isolated from its background, but the use of an optical device to project the image directly onto the canvas on which he was painting seemed unworkable according to Hockney‚s report.  It is hard, therefore, to integrate the optical precision with which Falco is supporting the hypothesis with the aversion to the use of optical aids described by Hockney as the artistic practitioner.

Text Box:

Lorenzo Lotto  őAngel of the Annunciation (c.1530)

We are left with the implication that artists relied on native skill and intensive practice to achieve their life-like effects.  Positive evidence for this view may be obtained from paintings where the subject is caught in motion.  Although easy enough to achieve with a camera, such depictions were a tour de force in the days before photography.  A striking example is available from the work of Lotto himself: his painting of the Angel of the Annunciation, who was often depicted at the moment of landing on earth, with garments aswirl. 


To attempt to set up this scene for optical projection with a model would clearly be impossible.  The figure is off-balance and the garments are clearly floating in the rush of wind.  Even if the model could have approximated this pose by leaning against a support, the diaphanous fabric would have hung limply by her sides.  The spectacular dynamism of the action captured by Lotto in this scene must have been purely a product of the artist‚s observational memory and painterly skill in rendering the fluidity and freedom of the fabric in motion.  In our age of high-speed photography, it is easy to forget that only the painter could capture time in flight in this way.  Lotto‚s picture illustrates that an ill-considered appeal to optical devices does disservice to the wide-ranging abilities of the Renaissance masters of Hockney‚s subtitle.



Thanks to David Stork and Amy Ione for comments on an earlier draft of this paper.




Baigre, Brian S.  (2000) őThe scientific life of the camera obscura‚ Optics and Photonics News, 11(2), 18-21.

Berenson, Bernard (1956) Lorenzo Lotto: Complete Edition. Phaidon: London.

Brown, David A., Humfrey, Peter & Lucco, Mauro (1997) Lorenzo Lotto: Rediscovered Master of the Renaissance. Yale University Press: New Haven, CT.

Edgerton, Samuel Y.(1991) The Heritage of Giotto‚s Geometry.  Cornell University Press: Ithaca.

Field, Judith V. (1997) The Invention of Infinity: Mathematics and Art in the Renaissance. Oxford University Press: Oxford, UK.

Hockney, David  (2001) Secret Knowledge: Rediscovering the Lost Techniques of the Old Masters. Thames & Hudson: New York.

Hockney, David & Falco, Charles (2000) Optical insights into Renaissance art.  Optics and Photonics News, 11(7), 52-59.

Kemp, Martin (1990) The Science of Art: Optical Themes in Western Art from Brunelleschi to Seurat. Yale University Press: New Haven, CT.

Liedtke, Walter (2001)  Vermeer and the Delft School.  Metropolitan Museum of Art: New York.

Lvova, Irina & Pavlova, Svetlana (1993) The Hermitage: Western European painting, 13th to 18th centuries. Aurora Art Publishers: St Petersburg.

Miller, Jonathan (1998)  On Reflection. National Gallery Publications: London.

Pirenne, Maurice H. (1970)  Optics, Painting and Photography. Cambridge University Press: London.

Wechsler, Lawrence (2000) őThe looking glass‚ The New Yorker, 31 January, 64-75.

[i] The further alternative that the tapestry itself did not have accurate geometry is highly implausible, based on this geometric analysis.  Although it is difficult to show directly, this alternative would imply, for example, that the octagonal motif on the tapestry itself would have to have been drastically distorted in order to account for the steep convergence of the construction lines.  While some misalignment is possible, such a pronounced deviation from parallel is highly unusual in Persian rug designs.