Artwatch UK

Posts tagged “Computer manipulated paintings

Good Science; Over-Reaching Science; Over-Promoted Science.

24 February 2014

On February 10th the Daily Telegraph published a letter from a professor of chemistry at University College London (Robin J. H. Clark) questioning the relationship between art and science in general terms and with regard to a supposed Chagall painting featured on a recent BBC Fake or Fortune television programme. Prof. Clark expressed particular concern over art world failures to heed the testimony of available scientific techniques.

In the late 1980s the UCL chemistry department had developed a non-invasive technique (“Raman microscopy”) for identifying both natural and synthetic pigments within paintings. Because the latter have known dates of invention, their presence in a picture can establish the earliest date at which it could have been produced. This technique is said by Prof. Clark to have been known to Sotheby’s by 1992. The Chagall painting, he pointed out, could have been exposed as a fake at any point in the last 20 years. He further reported that the painting was exposed as a forgery in his UCL laboratory in July last year in the presence of its owners and the presenters of Fake or Fortune:

“I am disappointed that neither of the presenters of Fake or Fortune made this clear. The conclusion that the painting is a forgery is based on our spectroscopic results, which showed that at least two of the key pigments had not been synthesized until the late Thirties, putting the earliest date for the painting at 1938, long after the supposed date of 1909-10.”

Because of the unequivocal nature of those technical findings, Prof. Clark (rightly) observed that the Chagall Committee in Paris, to which the painting was sent, had no option but to confirm the forgery. He also asked how art historians might be encouraged to read science journals so as be informed about “significant developments in science as applied to arts”. In part, his question is fair and urgent. The art market’s notorious governing trade dictum is caveat emptor (buyer beware) – while auctioneers and dealers may take every pain to verify their claims, it is ultimately for buyers to satisfy themselves that attributions and conditions are as described. Auctioneers can only submit works to (possibly disqualifying) technical analysis with owners’ permission. Dealers who buy at auctions almost invariably have works restored but are not required, when selling works on, to disclose which if any tests may have been run.

Support on the extent to which scientific (and also historical and visual) evidence is ignored or manipulated in the interests of “boosting financial rewards in attributing paintings to particular masters” was given in an Observer interview on February 23rd (“Revelealed: the art experts who pass fakes as authentic”) by Professor Martin Kemp, a Leonardo specialist. In the same report by Dalya Alberge, Nicholas Eastaugh, a leading independent scientist (of Art Access and Research), described the present climate as being both without standards and “totally unregulated. It’s a Wild West.”

However, much as we sympathised with Prof. Clark’s impatience with some art world practices, we could not endorse his call for a blanket acceptance of all scientific methods presently being applied to works of art. As we put it in a letter to the Daily Telegraph (published 12 February):

“Professor Robin Clark (letters February 10) calls for developments in science to be applied to art. If sound science is underused by the art trade, more questionable ‘scientific studies’ have been used for many years to offer assurances that picture-cleaners’ solvents have been a safe method of stripping varnishes and repaint from old pictures. As the current issue of the journal of the International Institute for Conservation of Historic and Artistic Works makes clear, the understanding in the art and museum world since the Sixties of how solvents work has been seriously flawed scientifically. Because important intermolecular interactions have been ignored, the theoretical model used cannot predict, as assumed, the actions of solvents on the underlying paints.”

History teaches that the many cumulative “scientific” defences of restorations have best been treated with scepticism. In 1977 Kenneth Clark admitted founding the National Gallery’s conservation science department precisely to bamboozle critics and dupe the public. In later years the Gallery pioneered a new mongrel discipline known as Technical Art History in which curators, conservators and conservation scientists pool expertises so as to arrive at some seemingly “scientifically underpinned” consensus on aesthetic decisions. In reality curators were glossing authority already-ceded to restorers. As the National Gallery restorer Helmut Ruhemann wrote in 1968: “Although the art historians in charge of pictures are officially responsible for the policies regarding cleaning, they naturally form their ideas in the first place from what they are told by their restorers.”

In its guides to conservation the National Gallery presently claims that while its restorations are carried out for aesthetic rather than conservation purposes, and while each restorer imposes a personal aesthetic taste on pictures, it considers all aesthetically various outcomes to be equally valid so long as they have been carried out “safely”. The contention that the (claimed) safety of cleaning methods can underwrite conflicting aesthetic outcomes is a non sequitur. Besides which, no claims have proved more unreliable than those of cleaning solvents’ safety.

The crucial and sometimes wilfully over-looked cultural truth is that there are no properly scientific means of comprehending art’s variously created aesthetic values and relationships. When reiterating this point in our post of 7 February 2014 (“From the Horse’s Mouth ~ Seventy years of worthless ‘science’ and reassurances on the safety of picture cleaning solvents”) we were able to disclose the most recent and most damning evidence of the un-soundness of past scientific endorsements of picture-cleaning solvents.

Notwithstanding these spectacular technical reverses, this month the press has been chocked with uncritical “Good News” accounts of scientific advances in the arts. Most newspapers and the BBC carried claims that scientists had “digitally reconstructed” the original appearance of a Renoir painting in which a former pink background had faded. By coincidence, this claimed miraculous virtual recovery had also been made by “a technique known as Surface-Enhanced Raman Spectroscopy (SERS)” carried out at the Art Institute of Chicago.

The BBC reported that “Northwestern University chemist Prof Richard Van Duyne pioneered SERS. He said the Renoir demonstrated why the Raman technique was fast becoming an invaluable tool for studying artworks: ‘You get tremendous information about the origins of the painting, the techniques of the artist, an understanding of the fading mechanism, and the ability to restore the painting.’” Note that speculative hypotheses are now being presented as sound platforms for restorations. In the art world it is frequently the dogs that don’t bark that matter most. Note that this wonder technique which addresses changes resulting from natural causes would seem to have no powers or potential with regard to the more common and much more seriously deleterious man-made changes made by restorers. Given that both types of injury are easily evident by eye to anyone lifing a picture out of its frame (see Figs. 2 and 3), the silence of “science” on the latter injuries can only seem self-compromising .

In a letter to the Times (February 17) we protested:

“The claim that scientists have recreated the original appearance of a Renoir painting (‘Laser technique shows masterpiece as Renoir intended’, Feb 14) is unfounded. All elements of a picture undergo natural changes over time. To these, further unnatural changes are added by restorers and their invasive paint-penetrating solvents. Compensating for a single faded pigment does not constitute a recovery of a picture’s original appearance. Rather, it offers a further falsification: a single artificially simulated ingredient within a remaining, generally altered and debilitated surviving whole.”

Our letter was accompanied by one from a Professor of Allergy and Clinical Immunology at Imperial College London, making a far-fetched claim that the fact that a synthetic red dye used in paintings had also helped in the discovery of an important white blood cell constituted an unusual “bridging [of] fine art and science”.

While Raman microscopy could certainly disprove the claimed date of the fake Chagall, it seriously misleads the public to present speculative and hypothetical digitally manipulated reconstructions as if literal recoveries of original conditions. On February 22nd the Economist reported an account of another digital re-mastering of real paintings delivered at this year’s meeting of the American Association for the Advancement of Science. The Economist too saw a bridging of the divide between art and science, which it likens to a resolution of the science/art schism of which the chemist and novelist C. P. Snow complained in his famous 1959 lecture “The Two Cultures”. The report also reveals, however, that what was presented as a recovery of the murals’ original conditions was in fact a double hypothetical reconstruction. Not only had Rothko’s colours faded, so too had those of the contemporary photographs of his murals that were to serve as the basis for a digital re-mastering of the actual paintings. Despite the methodologically dubious procedure of digitally re-mastering actual paintings on the back of digitally re-mastered photographs, there was customary breathless admiration for this latest claimed technical miracle:

“In the case of the Holyoke Centre’s Rothkos […e]ach had faded differently, depending on its original colours and how much sunlight it had seen. And various parts of individual paintings had faded at different rates, too. But modern technology allows optical illusions to be finely crafted indeed. The paintings are continuously observed by a high-resolution camera. Its images are compared, pixel by pixel, with the idealised versions provided by the restored photographs. A computer then works out, moment by moment, what mixture of light to shine back to make the faded originals match the vibrant reconstructions—with no messy repainting necessary. For now, the paintings remain under wraps while the museum at which they are stored is renovated. One day soon, though, they will be on display in all their illusory glory.”

There was no discussion of the consequences of viewers’ bodies blocking the projected “correcting” coloured lights. What we are witnessing in this heavily promoted technical bonanza is not a genuinely increased understanding of art by courtesy of scientific advances. If the attempt to increase public understanding of the degree to which even quite modern paintings have suffered alterations since their executions was a real ambition of museum staffs and conservation scientists, it would be imperative for them to discuss (and demonstrate) the largest single source of alterations and adulterations: “restoration” treatments. In the absence of such an agenda, what we see unfolding is a cultually diversionary Big Push by certain professional groups into new and uncontroversial employment pastures where the potential pickings and funding opportunities are immense – there is scarcely an old picture in existence where some pigments have not faded. This virtual remastering show is one that could run and run. But who might fund and who might execute research into all those paintings that suffered far more grievously from the chemical coshes of restorers?

The real problem in the arts is not an insufficiency of technical or scientific assistance. It is deeper and more fundamental. Its root lies within institutional withdrawals from exercising properly critical considerations. The non-appliance of due critical practices is long-standing. There were uncritical responses in the late 1990s when (as we reported in our first post) the National Gallery used a computer-manipulated photograph of an actual skull as the basis for a hypothetical virtual reconstruction of missing parts in Holbein’s “The Ambassadors” which led to the redrawing of Holbein’s skull in defiance (or ignorance) of the perspectival systems of the artist’s times. More recently, the Tate repainted large lost parts of a flood-damaged work on the basis of early colour photographs in the course of a “restoration”. In our uncritical, increasingly “virtual” cultural universe it is more urgent than ever that museum curators should return to acting primarily on sound scholarly appraisals and aesthetically informed insights, and that they should not further devolve their responsibilities to technicians who may or may not be properly alert to matters aesthetic and artistic.

Michael Daley

Comments may be left at: artwatch.uk@gmail.com

Printer-friendly PDF version of this article

Fig. 1: Above, top, Marc Chagall’s “Reclining Nude 1911” which is said to have been the source for the fake Chagall, “Nude 1909-1910” (above), as reproduced together in the Sunday Telegraph (2 February 2014).
An entire programme in the BBC’s Fake or Fortune series was spent examining the technical composition and the provenance of the fake version (which, incredibly, was dispatched to the Chagall Committee in Paris which not only declared the work a dud but threatens to have it destroyed) when a single glance at the two works should have been sufficient to establish that both cannot be by the same artist. Where that of 1911 displays a boldly deconstructing and reconstructing treatment of forms and spaces that is expansive and pictorially dynamic (as well as being massively indebted to Picasso’s then recent and revolutionary cubist works), the other is manifestly derivative and feebly handled, leaving the picture’s subject looking not so much set in a specially re-ordered non-Euclidian space, as pasted onto a monotonously and repetitively drawn and coloured theatrical back-cloth.
Above, Fig. 2: a detail of a Turner water-colour in the British Museum which had been protected from light damage at the left edge by the frame. (See plate 5 in the “Museum Environment”, 1986, Butterworth-Heinemann.)
Above, Fig. 3: a detail of Frans Hals’ “Banquet of the Officers the St. George militia company”, showing a strip of original green glazing that had been protected from restorers solvents by the frame.
Above, Fig. 4: the much reproduced Renoir, “Madame Léon Clapisson” (here as on the BBC) showing the painting in its present condition at the Chicago Art Institute on the left, and in an attempted digital reconstruction of its original (1883) condition on the right.
Above, Figs. 5, 6 and 7: details, top and centre, of “Madame Léon Clapisson” as found today, showing along the picture’s top edge a surving strip of an originally pink background achieved with a glaze of carmine lake, or cochineal, pigment. Scientists have used the investigative method known as “Surface-Enhanced Raman Spectroscopy (SERS)” in an attempt (above, at Fig. 7) to recreate the picture’s original appearance.
There has been no mention in any reports on this attempted reconstitution of some consideration having been given to changes in the painting that had occurred not as a result of exposure to light but as a result of exposure to restorers’ solvents, swabs and scalpels.
The painting itself and the virtual reconstruction is presently on exhibition at the Chicago Art Institute. The exhibition was supported by research funding provided by the Getty Foundation, the Grainger Foundation, the David and Mary Winton Green Research Fund, the Samuel H. Kress Foundation, and the Andrew W. Mellon Foundation. It is said that with “this new knowledge and new technologies such as nanotechnology, laser light, and advanced image processing software, the conservation department has been able to reconstruct the work’s original colors in a full-scale digital reproduction.”
A PIONEERING DIGITAL ATTEMPT TO RECOVER A PICTURE’S ORIGINAL CONDITION AT THE NATIONAL GALLERY (LONDON)
Above, Fig. 8: The National Gallery’s “The Conversion of the Magdalen” attributed to Pedro Campaña.
Above, Fig. 9: The near contemporary copy of the National Gallery’s “The Conversion of the Magdalen” that was made by Luca Longhi and is presently in the Villa Borghese Collection, Rome.
The National Gallery claims credit for pioneering the new collective discipline known as Technical Art History. A key weapon in its long, proselytizing campaign has been the publication since 1977 of an annual report dedicated to conservation activities – its Technical Bulletin. The issue of 2001 (Vol 22) carried an article “Colour change in The Conversion of the Magdalen attributed to Pedro Campaña” that was jointly authored by Marika Spring, Nicholas Penny, Raymond White and Martin Wyld. Spring and White were members the science department, Penny was a curator, and Wyld was the head of conservation. It was thus a textbook collaborative effort made under the rules of Technical Art History.
The combined expertises were brought to bear on a striking problem with the painting’s physical and optical conditions: there had been severe deteriorations in the colours of many of the draperies, not least in those of Christ. Many draperies were now brown or yellow-brown, where once they had been blue, green or red.
Microscopic samples were taken from some of the figures and analysed in an attempt to identify their pigments and to “investigate whether there was any peculiarity in the technique and the materials that could have caused such serious degradation.” Highly detailed examinations established that the blue pigment – smalt – had deteriorated; that a red lake pigment (likely containing dyestuff from the cochineal insect) had faded; and that green glazes containing copper had turned brown. None of these changes were remarkable in themselves, except, perhaps, in their extent.
What was remarkable was that an attempt was made to reconstruct the “altered colours by digital imaging”. It was explained that the changes which had destroyed the picture’s balance of colours had to be accepted as irreversible. Nonetheless, the attempt was made to gain some impression of the original appearance by manipulating a digital image of the painting – specifically, “by applying image-processing techniques”. Clearly, in such an exercise, the nature and type of image-processing software used would be of crucial methodological significance – how and in what manner was the base digital reproduction of the picture to be manipulated?
Explanation seemed to be to hand in a footnote [27]. Alas, it read flatly as follows: “The technical details of the process of reconstruction of the colours by image processing on the digital image will be described elsewhere.” No less disturbing than having to take the means and manner of the manipulations on trust, the account that followed of the factors of consideration suggested a Technical Art Historical methodology more Heath Robinson Contraption than Hi-Tech Sophistication.
Because the original colours no longer exist on the painting, some simulacrum of each had to be produced to feed into the image-manipulating software. Thus, “colourimetric measurements on painted-out samples matching the pigment mixtures and the layer structures were used as a reference.” Clearly, achieving a reliable point of colour reference was vital to the integrity of the exercise. But how reliable were the painted-out samples? Not very, it seemed on the authors’ own account:
“For the smalt and red lake pigments this posed some problems. Smalt manufactured to a nineteenth-century recipe is available today, but contains a higher percentage of cobalt than than smalt in sixteenth-century paintings and none of the impurities that are commonly found in the glass.” Notwithstanding these departures from the original materials used on the painting, this smalt was used for the base references. Because the modern smalt is much stronger in colour than that of the sixteenth-century, an attempt was made to correct (lessen) its force by adulterating it with “finely ground alumina” in attempt to “to try to simulate the colour of the sixteenth century smalt”. Confidence in this adjustment was not high because “this is a difficult judgement to make, since in paintings of the period smalt has always degraded.” Had the painting been a seventeenth-century work the exercise would have been easier because by then the smalt was commonly mixed with lead white pigment, which afforded some protection. Even though this work was not of the seventeenth-century, samples from that period were used a guide reference in the digital manipulations.
Establishing a reference point for the original lake pigments was no less problematic: “Comparison with the deep shadows on Christ’s red robe, which retain their red colour, made it clear that the hue of the test plate was more purple than the red lake in the painting…” And what of the outcome of this, at best, approximate method?
Above, left, Fig. 10a: the computer-manipulated attempt to recover the original colours of Christ’s draperies.
Above, right, Fig. 10b: a detail of the Borghese Villa copy shown above at Fig. 9.
It probably goes without saying that the figure of Christ seen at Fig. 10a seems a most implausible reconstruction. It is claimed by the authors, however, that: “The deeply saturated colours which replace the deteriorated brown, although rather flat because of the loss of the modelling which cannot be reconstructed, balance well with the well-preserved draperies painted with vermilion and ultramarine.” Given that, on the authors’ own admission, the simulated blues and reds are significantly different and more intense pigments, how credible can this claimed correspondence of colours seem?
The article concludes on an assertive note of self-satisfaction: “The detailed technical examination of the ‘Conversion of the Magdalen’, and the process of reconstruction of the colours in the digital image, has produced some deeper insight into how the deterioration of pigments has affected the colours in the painting.”
This was followed by a claim that is quite remarkably at odds with the visual evidence presented (see Figs. 10a and 10b): “Although the strong and deep colours of the reconstruction initially seemed rather startling, they receive strong support from comparison with the Borghese version of the painting [shown here at Figs. 9 and 10b] – which is especially gratifying since the reconstruction was made before the transparency of the Borghese version was available to us.” Given that the Borghese version is on all accounts markedly better preserved that the London picture, what might explain the former’s richer, warmer red drapery and darker, more sombre blue drapery?
Although the authors express themselves as being satisfied with the accuracy of the reconstructed colours, they do concede other problems: “The reconstruction is not, of course, an accurate portrayal of the original appearance of the painting – the lost modelling in some of the draperies cannot be recreated…”
Thus, we see that this exercise has been directed at a single component part of the painting – its self-contained areas of local colours – and that, in the execution, that part has been wrenched from any relationship with the picture’s tones, shading and modelling. This severance is painfully evident in the comparison at Figs. 10a and 10b. It would beggar belief that the National Gallery’s experts could see any sort of vindication for their efforts in the Borghese version were it not for that institution’s by now too-deeply ingrained to be recognised tradition of pursuing autonomously bright colours during restorations at the expense of form and pictorial coherence. Not only are the colours of the Borghese drapery more sombre and chromatically integrated – and jointly more skilfully integrated with the plastic values – but we see also in the National Gallery picture a characteristic debilitating weakness of modelling in the too-brightly scrubbed surfaces of the flesh areas. (It is depressing beyond belief that our national pictorial vice should recently have crossed the English Channel and now be menacing Leonardos at the Louvre.)
It might be contended that we are not comparing like with like. As the authors point out, the one work is a not an altogether strict copy of the other. Moreover, the Borghese version is acknowledged to be in superior condition: “the better condition of areas painted in red lake in the Borghese painting is strong evidence that it has not been subjected to such harsh environmental conditions as the National Gallery painting…The Borghese picture has spent almost all its life in two collections in the same city, whereas, the National Gallery’s picture has belonged to at least half a dozen collections and has passed on at least three occasions through the art trade, but too little is known about the conservation history of these paintings, and the conditions in which they have been kept, to explain the difference in preservation.” The euphemistic use of the term “environmental” in lieu of “restorational” and the sly allusion to possible bad restoration experiences at the hands of the “art trade” cannot gainsay the fact that there is abundant evidence of works held at and restored within the National Gallery suffering catastrophic losses in the course of a single in-house restoration – as the before restoration (left) and after restoration (right) comparative details of Rubens’ portrait of Susannah Lunden (shown below at Figs. 11a and 11b) testify.
Click on the images above for larger versions. NOTE: zooming requires the Adobe Flash Plug-in.


wibble!