Vermeer-related lecture

February 28th, 2015

Philip Steadman
Darwin Lecture Theatre, Darwin Building, London
March 5, 2015, 13:15-13:55
price: free
contact: +44 (0)20 3108 3841 |
event page

In 2001 Philip Steadman published Vermeer’s Camera, a book that offered new evidence that the great Dutch painter relied on optical methods. An American video engineer Tim Jenison read the book and, believing he could take the argument further, proposed a simple arrangement of lens and mirrors that Vermeer might have employed. Jenison used this setup to paint a version of Vermeer’s Music Lesson in the Queen’s collection. The process was filmed for the Oscar-shortlisted documentary Tim’s Vermeer, released in 2014. Jenison’s method throws more light, literally, on how Vermeer could have achieved his distinctively “photographic” tonal effects.

The lecture will be streamed live online and recorded for YouTube or downloaded.

35 Responses to “Vermeer-related lecture”

  1. craig


    Have you had a chance to see the lecture? I’m curious to hear your impressions. It seemed to me that Steadman did as good a job introducing the subject as one could hope for, considering the limitations imposed by the format and time. It was a lot of ground to cover in just 45 minutes.

    Watching the lecture made me realize how spoiled we were by the discussion that took place in the comments section of your “Tim’s Vermeer… from a painter’s point of view” post. It could be a bit freewheeling at times, but overall it was a great forum to explore in some detail the ideas raised by Tim’s work. I was really impressed by a lot of the comments. And of course, we were fortunate to have Tim involved in the conversation. It was helpful to hear him explain aspects of his experiment that didn’t fit into the film.

    I’ve looked, but have been unable to find anything about Tim’s experiment that is remotely as thorough and informative as the discussion you hosted. Steadman’s lecture got me wondering if it might be worth returning to the subject here.

    I think everyone agrees that when the earlier discussion ended, it did so for the right reasons. Facing the loss of a parent is a uniquely awful experience, and my heart goes out to Tim. I don’t know if he will ever read this, but if he does, I hope he will accept my sincere condolences.

    The prospect of renewing the conversation without Tim is problematic. Not only was his personal presence welcome, but his contributions were invaluable. I’m not sure we can come to a clearer understanding of unresolved issues (like the tonal map of the Music Lesson for exp.) without his input. Still, I think there may be some value in trying. What do you think?

  2. Jonathan Janson


    Thanks for the comment.

    Yes, I followed Steadman’s lecture in streaming. It was, as you say, about as good an introductory course to the issues as one could desire. Perhaps you were able to sense Steadman’s lucidity? Really, the main progress in Vermeer’ studies in the last half of the 20th century were largely due to Montias and Steadman, oddly, both outsiders to the art history enclave.

    As far as the blog’s post goes, I agree with you on all counts. I found nothing else similar so far and it was a pleasure to have started it going. The only thing that lacked was Steadman, no? I had hoped Tim would write something “conclusive” after, but as far as I can tell that does not seem to be in the cards.

    Personally, I urged Tim to backtrack a bit and do some controlled and systematic experiments. I would like to see different painters use the device and see how close they could all get to a single, simple motif (like Vermeer’s wall with just the window, pavement and a black-framed picture on the wall). How really good are painters at capturing Tim’s “impossible” tonal gradations? One could have them do it, perhaps with and without, the device. I feel if the experiments were devised intelligently, with objective criteria and temrs of comparison, they would reveal much. I suspect that painters would better than what Tim believes but that is really not the point. I think we could all learn immensely IF we could get on the same page. The whole problem of perceptual constancies, tone, optics and painting is one that has barely been scratched and yet it is at the base of so much pictorial discourse. And this speaks greatly of art history’s limits. I admit, it would be a titanic endeavor for anyone, but one that could better the appreciation of the “invisible” aspects of painting.

    BTW, to be truthful, Vermeer studies have slowed down in the last 10 years or so. We may be no longer able to make progress.

    I am currently at work on a 400+page revision of my (Ebook format) studyon Vermeer’s technique and I’ll throw that out for discussion fairly soon. It took one year to write and a whole year to revise. This time, it’s heavily illustrated.

    Let me see if I can find a real good question for another post.


  3. craig


    Thanks for your reply. I think you touched on what was a key point of confusion for me in trying to evaluate Tim’s experiment. It seemed like it was missing ‘objective criteria and terms of comparison.’ The claims about “impossible” tones were particularly tough to pin down.

    Tim argued that Vermeer got the values right in a way that the unaided eye could not, making it likely that he used a comparator device. As you suggest, one rational way to test this would be to have skilled painters paint the same scene with and without a comparator. At a minimum, this would help give some sense of what the parameters of tonal accuracy might be. I also think Tim’s ambitious approach of recreating the Music Lesson was a good idea. It’s a classic attempt to “replicate the circumstances, then duplicate the results.” If the tonal maps match, then this could be seen as evidence that Vermeer used a comparator.

    But as I understand it, if we were in the room depicted in the Music Lesson, using a comparator device and paint to record the falloff of light across the back wall, the results would not match the falloff of light in Vermeer’s painting. If the wall surface was historically accurate, whitewashed plaster, the overall tone of the wall in our painting would also be noticeably brighter than the wall in Vermeer’s. None of this bodes well for Tim’s theory.

    Tim’s painting looks the way it does because he made two important modifications to his version of that room. First, he tinted the plaster, darkening its overall tone. Then he added shades outside the upper windows to alter the flow of light across the back wall. Even with these changes, the tonal map of his painting has noteworthy differences with the Music Lesson.

    So the problem I have is that I can’t square Tim’s claims about Vermeer’s tonal accuracy with the results of his experiment. I don’t understand what his standards for determining accuracy are.

    And I don’t understand his methodology. It seems like he’s working backward from his conclusion. He starts with the premise that Vermeer’s tone is accurate, discovers in the computer modelling phase that the falloff curve on the back wall doesn’t really resemble the painting unless he puts translucent shades over the upper windows and assumes that Vermeer must therefore have used translucent shades. Maybe he really did stumble upon Vermeer’s method for altering the lighting conditions in his studio, but if he’s going to talk about the limitations of the retina and matching Vermeer’s falloff of light, he has a responsibility to mention those shades every time he does so.

    I hope Tim took your advice to heart and has been working on some more systematic experiments. And I hope that when the time is right, he will share the results.

    The revision to your book on Vermeer’s technique sounds like a huge project. I’m looking forward to hearing more about it.

    In the meantime, if it isn’t too presumptuous, could I offer a suggestion for a post? Back in September, Tim posted a computer model of the room in the Glass of Wine. He said “the window in the foreground is replaced by a virtual piece of glass the same size and shape as the window in the painting. If you look closely you can see a reflection of the chair in the glass.
    The reflection of daylight off the glass is creating the brightly lit patch just below the window. The window light is striking the bench as in the painting, although not as dramatically.”

    So Tim was able to achieve a reasonable approximation of the lighting by assuming the window was the size and shape depicted in the painting. But in Steadman’s reconstruction, the window casement is much larger. Steadman suggested the painting may have been a collage of two projections. If it is two collaged projections, how do we get the reflection of the chair in the glass or the bright spot under the window? If the geometry of the window in Tim’s reconstruction is right (as it would have to be for the lighting scheme to work), then what about the geometry of Steadman’s reconstruction?

    I’ve been curious about this ever since Tim posted the image, but at the time the conversation was going in another direction. Seems like it might be worth considering now.

  4. WWick

    Hi Craig and Jonathan,

    I also hope we can continue the discussion, though I won’t be able to devote as much time as I’d like in the immediate future. Jonathan, I wonder if there’s a way to allow commenters on your site to edit past posts. I had a bad run of typos, and the last I looked there were a bunch duplicate posts due to some unknown problem. I’d be happy to go back and clean up some of that.

    Anyway, I agree that Steadman’s summation of the problem was masterful, but so too was his book in that regard. I don’t think he added many new insights regarding Tim’s experiments, and like you, Craig, I have a difficult time with the lack of precision over exactly what is meant by accuracy of tone. As we have found, there would have to be a lot more faux painting and window shading for Tim’s methods to fully capture the emotional resonance of Vermeer’s treatment. It was particularly disconcerting that Steadman introduced his audience to the Music Lesson with a sickly, sepia toned scan, then switched to a better rendition for side by side comparisons. Not that there was anything nefarious going on, it’s just that it undermined the discussion at hand.

    I would add that there is also a lack of precision over what is meant by painting from “reality” versus painting from the imagination. As Steadman marched through his comparison of “real” objects versus props within Vermeer’s painting (the fidelity of which were well within the grasp of conventional painting methods), I took issue with the suggestion that there was no “imagining” involved – particular with regards to floor tiles, wall tones, instrument color, and the mixing and matching of different lighting – all effects for which there is strong evidence a lot of fudging occured, and possibly some pure fiction too.

    I have no doubt Tim has succeeded in making a “transcription machine”, but he has yet to succeed in building and lighting a set that his machine can render with the full emotional impact of Vermeer’s original. If we are to concede what Tim has done is close enough, do we judge all art by similarly loose standards?

  5. craig

    W Wick,

    Great to hear from you again. As usual, I find myself agreeing with what you wrote. Though in this case, noting the agreement, I’m also reminded of the problems we face in moving forward without Tim. You, Jonathan and I seem to be on the same page regarding Tim’s experiment and what it does and doesn’t say about tone, but I expect Tim disagrees with us. Without his input, we run the risk of a very lopsided and incomplete evaluation of the issues involved. Tim may have additional information about his experiment that could sway opinion in his favor. I’m not sure how we get around this. I know that he offered to make all of the footage and data from his experiment available. Did he? If so, maybe someone who has reviewed it can be persuaded to contribute to the conversation.

    As for Steadman’s lecture, I had the same reaction to his treatment of ‘reality’ vs. ‘imagination.’ It seemed like he was responding to a caricature of Liedtke’s argument, rather than dealing with it seriously. I chalked it up to using overly broad strokes to fit the time allotted. It may be the sort of argument that’s better hashed out in a forum like this.

    The thing that really jumped out at me was what he said about the ceiling beams. He showed the three paintings with a visible ceiling, noting that in each case the ceiling seems to be about the same height. But that’s not true, is it? According to the measurements in the appendix of my copy of Steadman’s book, the ceiling in the Music Lesson is over 20 inches higher than in the Art of Painting and just shy of two feet higher than the ceiling in the Allegory of Faith. I certainly wouldn’t characterize all three as “about the same height.” The two allegory paintings maybe, but the Music Lesson is quite the outlier.

    This gets to the heart of what has always confused me about Steadman’s book. Maybe I overlooked or have forgotten a key passage, but as I understood it, Steadman used his geometric reconstruction of the Music Lesson as the basis for his reconstruction of Vermeer’s studio. Is that right? According to Steadman’s measurements, the windows in the Music Lesson are significantly larger than those in the other five paintings. Those five had window sizes similar to each other, but it’s the outlier, with its much larger windows that seems to have served as the blueprint for the apertures in the side wall. This might explain why the windows are too big in the photo reconstructions of the other paintings. That always struck me as strange, but I didn’t know what to make of it.

    Tim’s computer model of the Glass of Wine seemed to put pressure on exactly that point. I’m probably missing something, but it seems both the lighting scheme and geometry of the foreground window work if one assumes the smaller windows/embrasures. Neither works in Steadman’s model. I’m hoping you, Jonathan or someone else might have some insight into how to resolve this conundrum.

  6. Les Blatt

    I’ve had a problem with Tim’s approach from the start, although I do appreciate the wonderful optical comparison instrument he developed. What evidence is there that Vermeer actually tried to make a literal copy on canvas of the physical scene in front of him? When we marvel at the tonal qualities, why do we have to imagine that those were the tonal qualities of a wall in Vermeer’s studio, rather than an effect he wanted to achieve due to his own esthetic sense? Tim built a room that was a literal projection of the scene Vermeer created, and then showed that he could make some reasonable (but rather mechanical) copy of that room on his own canvas. A copy of a copy of a poster of Vermeer’s canvas.

    I am a scientist and a lover of Vermeer’s work, so I do not disrespect anything that’s been done to explicate either the optics or the art. I think Steadman has made a great contribution to the understanding of the relationship between the optical science and the art of Vermeer’s day. But I think the basic assumption of “Tim’s Vermeer” is a flawed one.

  7. craig


    I don’t know if you’re still checking in on this thread, but I hope you are, because you made some good points that are worth addressing. I think you’re right to call attention to the basic assumption of “Tim’s Vermeer.” From my point of view, this gets to the heart of where Tim’s experiment goes wrong.

    Instead of using the comparator to test the claims about Vermeer’s tonal accuracy, Tim started with the assumption that Vermeer’s tone was accurate and then modified his experiment to confirm his belief (see: selective shading, tinted plaster).

    In the film (as well as the articles and interviews that promoted it), Tim didn’t just claim that Vermeer’s tone was accurate, he argued that Vermeer got the tones right in a way that the unaided eye could not. When he said this, he wasn’t talking about Vermeer’s tone ‘in general.’ He was talking specifically about the falloff of light across the back wall – the very thing that Tim deliberately altered in his experiment.

    Maybe I’m crazy, but I think that’s a really big deal.

    Also, if anyone reading this knows of an article or interview in which Tim, Penn or Teller discuss how the shades and tinted plaster affected Tim’s experiment, could you please post a link below? I’ve been looking but haven’t been able to find anything.

  8. Jonathan Janson


    The “big deal” you mention arises only if one maintains that Tim experiments provides proof rather than a hypothesis. Perhaps the idea of a proof, or at least some pretty seriously flirting with it on the part of some, and the general tenor or the film has kept art historians from inquiring further to this exceptionally interesting topic.

    Another matter is that there is no way of extending Tim’s methods to any other painting by Vermeer, obviously not Tim’s fault. This places weight the whole hypothesis on a single work, and that’s pretty dangerous. And as we all know, Steadman’s hypothesis was accepted because it is based on a composite of measureable evidence.

    Moreover, I have the impression that Vermeer played quite a bit with value and that there is no single “method” that links all the paintings. For example, to my eyes, the mass shadow of the blue morning jacket of the Woman in Blue Reading a Letter looks all too dark, like a hole in the painting. Can we really go from light powder blue on the front side to almost black in the mass shadow in such a brilliant ambient light? Were there no reflections on the shiny material, no light bouncing off the walls and floors? Even the opaque gown shows a few tonal variances. The shadow is almost as dark as the pure black hanging rod of the map behind the figure (impossible if I am not mistaken). This anomaly is also evident when the garment’s dominant chiaroscural contrast is compared to the gown’s. Although the tonal value of the actual gown’s local color would seem to be slightly darker than that of the jacket, the value of its core shadow is lighter than that of the jacket, instead of darker.

    To my mind, Vermeer was a complicated man. He threw in everything, including the kitchen sink, to get the image he wanted. Theoretically, there is more than enough room in his approach for a comparator mirror in his game, but no definitive proof as of yet.

  9. craig


    Thanks for the reply. Respectfully, I don’t see proof as the issue here. I never expected Tim’s experiment to give definitive proof that Vermeer used a comparator, and I agree with him that in evaluating the evidence, “it’s a matter of likelihood not proof.”

    Tim said that ‘tone’ is what’s different about Vermeer and it’s why he did this experiment. Vermeer’s tonal accuracy is central to Tim’s theory, and while it can’t prove that Vermeer used a comparator, it was presented as increasing the likelihood that he did. For me, the ‘big deal’ is that the evidence from Tim’s own experiment argues against the hypothesis that Vermeer painted with tonal accuracy.

    Looking at all the information we have about Tim’s experiment (including Wick’s GIF of 11/3), Tim’s interpretation isn’t impossible, but it involves a series of very big assumptions. A simpler – and I argue ‘more likely’ – explanation for what we learned about tone from Tim’s experiment is that Vermeer painted a convincing illusion which appears tonally accurate, but isn’t.

    Even if one is inclined to believe that Vermeer used tinted plaster and/or temporary shades to achieve a more desirable lighting effect in his studio and that he then used the comparator to match tones in some parts of the painting but not others, surely it must be conceded that such would be – at best – an incredibly shaky foundation upon which to build the claim that Vermeer accurately recorded tones that the eye can’t see.

    While I think Tim’s argument about Vermeer’s tone is a bad one, I think I can understand how he was drawn to it. After all, it was while he was thinking about how Vermeer might have achieved his tone that Tim came up with the idea for the comparator device. The desire to maintain a connection to the source of his inspiration strikes me as perfectly natural.

    I should also say I think the comparator is a really cool invention. Tim absolutely deserves our respect and admiration for the ingenuity and determination it took to conceive and develop it. He’s invented an incredibly clever means of using optics in painting.

    He’s also done a good job explaining how versatile an instrument it is. But this versatility is precisely what makes it difficult to evaluate claims for its use in the past. If an experimenter grants himself the latitude to claim that an artist used temporary shades in a visible part of the tableaux, or for ‘dodging and burning,’ moved the lens to change the size of certain projected objects, and/or used the comparator in ‘this’ part of a painting but not ‘that,’ then he can argue he’s found evidence for the comparator in pretty much any realist painting he desires. It’s a methodology which guarantees an astonishing number of false positives, and offers no objective means for weeding them out.

    The comparator idea is a compelling one that deserves further testing, but I really don’t know how one could test for its use in the past in a way that avoids confirmation bias.

    Also, I don’t know what motivates art historians, but if they are generally reluctant to weigh in on Tim’s work, it may be that they’re waiting for someone with a background in science an optics to review all of Tim’s data.

  10. craig


    When I woke up yesterday morning, I re-read the message I had posted before going to bed and was horrified to discover what an incoherent mess it was. I want to apologize first for submitting that post before taking the time to organize my thoughts and second for what I’m about to do, which is take another crack at it. I hope it’s at least somewhat coherent this time…

    I see no problem with using Tim’s claims about Vermeer’s tonal accuracy as a working assumption when testing the possibility that Vermeer used a comparator device. Since the comparator works by a direct matching of paint color to light color, it’s reasonable to assume that a painter who used such a device would achieve the high level of tonal accuracy that Tim describes. Unfortunately, the moment Tim’s computer model demonstrated that the falloff of light across the back wall of the Music Lesson didn’t match the natural lighting conditions of the room, that working assumption was shown to be false, at least as applied to the that painting.

    The appropriate response to this situation is to report the discovery and consider alternate testing approaches. Instead of revisiting all the choices Tim made after this point, I will simply say that despite his best efforts, the results of his experiment undermine any argument for ‘objective’ tonal accuracy in the Music Lesson. Maybe I’m alone in thinking this is a big deal, but at a minimum, it’s a point that should be openly acknowledged.

    Part of the narrative surrounding Tim’s experiment is the idea that art historians are reluctant to accept the evidence of “crass rationalists” who bring a science-minded approach to the question of optics in painting. But Tim’s experiment wasn’t the example of clear-eyed scientific rigor it appeared to be at first blush. As I see it, Tim’s approach suffers from many of the problems that Dr. Stork found when evaluating the Hockney-Falco theory:

    “It is an intriguing and captivating theory, and before scholars had a chance to study the theory it was highly promoted in the public arena. Unfortunately, as stated, it isn’t a scientific or testable theory because proponents give no unique and objective way to tell which “certain elements” in a painting would have been traced and which not. As such, proponents change and retreat from their claims as independent scholars prove these claims about “certain elements” incorrect—changes that are always ad hoc (a Renaissance painter would have traced this passage, but not that passage) so as to try to salvage their theory. This is the hallmark of a failed or even unfalsifiable theory: much as Wolfgang Pauli described one physics paper, ‘That’s not right. That’s not even wrong.’” –Stork.

    From Wikipedia:
    “In science and philosophy, ad hoc means the addition of extraneous hypotheses to a theory to save it from being falsified. ad hoc hypotheses compensate for anomalies not anticipated by the theory in its unmodified form. Scientists are often skeptical of theories that rely on frequent, unsupported adjustments to sustain them. ad hoc hypotheses are often characteristic of pseudoscientific subjects such as homeopathy.”

    The problem I’ve been trying to describe isn’t that Tim failed to prove his hypothesis that Vermeer used a comparator, it’s that he ran a bad experiment.

    Moving forward, whoever takes up the task of designing new experiments to test the comparator idea has their work cut out for them. Since the comparator is a color/tone transcribing machine, how does one make the case that a comparator was likely used if the tones don’t match? If the idea being tested is that the painter used the comparator in just part of the painting, how does one test that without falling victim to the temptation of making the experiment fit a desired outcome? The comparator is a wonderfully versatile machine. If Vermeer was aware of it, that may have been a big part of its appeal. Unfortunately, today that versatility makes it tough to test whether or not a painter used such a device in the past. I can imagine a million experiments which show it’s ‘not impossible’ a given painter used a comparator. Coming up with one that demonstrates it’s ‘likely?’ That’ll be tough.

    I think the art historians are smart to wait.

  11. Jonathan Janson


    Sorry I did not pick up on the incoherencies of your first post, it looked ok to me. And I accept your criticism of my previous post. Thanks for showing me the point.

    Your second post is pretty robustly argued.

    To me, it’s fairlyimpossible to produce any sort of useful “scientific” data of the kind needed for Tim’s experiment with 350 year old pictures, for various reasons.

    First, we don’t know if they look the same way today as they originally did. I have stressed this point many times but no one has ever followed up on it. Colors change, and things get darker and lighter and retouched, not just a bit… but DRAMATICALLY. We simply don’t know if Vermeer’s Music Lesson reflects Vermeer’s original intentions, and when it’s a matter of calibrating tonal values, it’s a make-or-break argument. Looking at the Music Lesson, there seems to be many strange things going on. We should not ASK but KNOW if the wall in Vermeer’s Music Lesson is really how it was painted before preceding. It’s got ultramarine blue in it and ultramarine is treacherous. Look at the Delft tiles that run around the walls of A Lady Seated at a Virginal. Under the virginal they are actually lighter that the shadowed white-washed wall itself.

    Two, we don’t know what the original scene looked like. Can you test Vermeer’s tones if you don’t know what the lighting was in the room or if the colors of the original objects were in the first place? Or if they were all assembled in the same environment at the same time? As suggestive of reality as Vermeer’s values are, we do not know how much was invented and how much was edited. As a painter, I suspect that there is a huge margin of perceptual tolerance to erroneous tonal values and that the painter employed symultanously many different tactics to the the “look” of lifelikeness.

    Three, basing an argument on partial evidence found in only one painting is perilous, at the very least. The strength of Steadman’s arguments is that there exists measureable, verifiable evidence in a significant number of different paintings.

    Tim’s experiments are admirable ingenuity and in trying to nail down things with numbers, and he discovered many interesting phenomena that the art historical community blissfully ignores. But I am not sure what it all cooks up to. My current position remains that as far as the picture making process goes (or hopefully what we know of it), nothing I see in Vermeer’s paintings can prove the comparator was not used. But, alas, that’s hardly evidence that it was.

  12. Paul Bartholomeusz

    Guys does it matter whether he used a comparator, or a camera, or a compass for that matter? A million people could paint by numbers a million times without any one getting anywhere near what Vermeer does. Enlightens, and delights.

  13. Jonathan Janson


    Thanks for the comment.

    I think your “a million people could paint by numbers a million times without any one getting anywhere near what Vermeer does” is absolutely correct. To my dismay, during 40+ years of attempts to capture in my own work a bit of the je ne sais quoi quality of Vermeer’s art (using everything means possible except the kitchen sink)…I have failed miserably.

    However, the discussion on this blog, I believe, does not involve the question if the methods or machines that Vermeer presumably used could enable someone else to make paintings like Vermeer, but if Vermeer used them to make his. A Stradivarius makes beautiful music only to for who know how to play it. For all others it just screeches. The same would be with a comparator, or a camera obscura.

    I think Tim would gladly acknowledge that if he were charged to create an interior painting, say of his own house, which is NOT based directly on a Vermeer painting, it would probably look much much less Vermeer-like than his own Music Lesson. The shape of his shadows, the luminosity of his colors, the beauty of his paint and brushtrokes, the organization of his composition, the poses of his characters, the direction of light, to say nothing of the overall Gestalt, would all be wrong, just as wrong as those of my “Vermeer” paintings even though I have looked at Vermeer for many more years than Tim. In fact, even if Tim were to obtain the most accurate tonal values imaginable in his scene, even more accurately that those of Vermeer, they would not bring his work expressively closer to any of Vermeer’s. This is because, Vermeer’s painting is not just tonal accuracy, but a unique composite of all those things that millions of people could try for millions of years to duplicate but never really come close to replicating.

    Obviously, Tim makes no pretension to have painted a Vermeer, but only to legitimately demonstrate that Vermeer might have used a comparator mirror to help him paint, so what I have just written in no way debases “his” Vermeer or his experiments. There are those who believe a device such as the comparator was necessary to achieve the taonl accuracy exhibited in Vermeer’s paintings, and there are those who believe he achieved them through combined use of the camera obscura, a considerable dose of skill, conventional pictorial schemata…and last but not least, genius.

  14. craig


    I have a vivid memory of being in Brice Hobbs freshman drawing class and watching him with a pained look on his face say “Sometimes on my way home, I remember something I said in class, and it makes me want to run and hide under a tree.” That memory sticks with me because I feel that way all the time. My long-suffering friends would be quick to tell you that in conversation, I tend to babble. Actually, they’d be more like to say it takes forever for me to get to the #%&*! point. And by the time I do, they’ve lost track of what it is I was talking about.

    Ideally, sitting down to write offers the opportunity to clear away the clutter. I’m embarrassed to say that it’s usually a long, awkward process requiring multiple revisions. And it doesn’t always work out. The words still come out wrong. I worry also that in this attempt to whittle down to a clear point, I end up with something that comes across as too terse or abrupt, with a tone that doesn’t fit my intention.

    For example, I didn’t mean to criticize your post. I thought my lack of clarity had caused some confusion and I was trying to correct it. I’m sorry it didn’t come out that way. To be honest, I can’t think of anything you’ve written on this subject with which I disagree.

    I wholeheartedly agree with your point about how the paintings have changed over time and should have said so when you mentioned it earlier. If this conversation had taken place over a beer in Delft as cobalt once suggested, you would have seen me across the table, nodding vigorously in agreement, silent because I have nothing useful to add.

    Your second point is also well taken. Though I do think it’s possible to run experiments that shed some light (so to speak) on the subject. For example, Steadman’s photographic reconstruction of the Music Lesson helped to correct some erroneous assumptions about the angles of shadows. Also, I read an article a while ago that said Dr. Stork has been doing work with computer imaging/reconstructions of paintings and publishing his findings in peer reviewed journals. I think the reconstructions are probably like the one Tim worked up for the Glass of Wine. I don’t know if any firm conclusions have been drawn from that work. My guess is that the wording goes something like “consistent with” or “inconsistent with.” It strikes me as unwise to go beyond that for all the reasons you describe.

    I agree with your third point as well. My personal feeling is that Tim’s project was so flawed that it’s best to just start over, but we wouldn’t have to start from scratch. His warehouse/experiment room seems to me an ideal laboratory for experts in optics to design and run various experiments related to each of Vermeer’s paintings. They would need to collect new props to ‘dress’ each set piece, but a careful approach may yield measurable, verifiable evidence. If it were up to me, I’d have Stork and James Elkins co-direct the project.

    My current position on this subject is identical to yours, which I think you explained quite lucidly in your original “…from a painter’s point of view” post. If there is a difference between us on this, it may just be one of emphasis. I’m all for continued testing of the comparator idea, but I strongly believe that before that can happen there must be a clear, thorough and very public review of Tim’s data and conclusions. We need to know for certain whether his conclusions were drawn from the evidence, or if it was the other way around.

  15. Jonathan Janson


    Criticism or not, no harm done, I enjoy other people’s ideas.

    I was impressed by Hobbs too. I graduated in 72.

    And I thoroughly sympathize with your writing plight because I have been dealing with it intensely for the last year of reediting my book on Vermeer’s technique. As you know too well, painting technique is largely an unwritten story, so I must balance my professional experience, intuition, scientific results, naked eye observation and simple logic to flesh out some very complex issues, at all times attempting to make clear to the reader (and to myself) which is which. The chapters on modeling and composition have driven me to near mad. The result is that I walk around all day wanting to hide under a tree. But I figure someone has to start the ball rolling so others can read and critique.

    I am not too sure what experiments Tim can do to test his hypothesis but I can think of many that might help us clear up how well, or how bad, accomplished painters are at hitting tonal values on the head. I imagine more than Tim would suspect, but I would not be surprised if this was not the case. Strip the room of all the props except for the framed mirror and picture to the right, give the painter a canvas with a thin outline drawing and then let him try to get as close to the true tones as possible. This could be compared to a picture made with the comparator. I have a feeling that Vermeer used the black frames to help him gauge values like Cezanne used a black top hat. Note that the two broadest walls, where the falloff of light is most crucial (The Music Lesson and The Concert) both have two black objects that break the gradual shift of value from left to right. The first thing you do is to knock in the two frames with straight black and afterwards, it’s much easier to see where you are value-wise. The falloff of light in the Woman in Blue is easy, that is, it’s essentially one tone to the left of the figure and one to the right. A number of combinations of tone would furnish a reasonably account of light because there could be may different intensities of light and we do not know which one Vermeer originally painted from.

    I doubt if there are many players who are interested in reviewing of Tim’s data and conclusions. He’s neither an art historian nor Hockney.

    Yes, Stork and James Elkins might be the right people. Paul Taylor is a good name too. He is open but scrupulous.

    In any case, you do a very good job of paring down your ideas. I could never have found a better way to get them across, and they have gotten me to do some serious thinking.


  16. wwick


    While it’s true we don’t know what Vermeer’s painting originally looked like, we praise his art for how it looks today. To a large extent, we have to judge Tim’s experiment by that standard. I would add that that it might be safe to assume, that what we see now of Vermeer’s lighting patterns is likely representative of his intentions (unless the entire painting could be shown to be overpainted), even as his hues may have shifted. As to the hue, even though the stability of ultramarine and other pigments may be problematic, there’s still a question to why Tim’s studio set-up did not demand much in the way of ultramarine, as far as I know, and why he did not achieve the golden warmth of Vermeer’s instruments, even though he re-painted the instruments themselves in an attempt to do just that.

    I’m not a conservator, but common sense tells me that if Tim’s painting were to undergo an aging process for over the next 350 years under conditions similar to Vermeer’s original, it would not magically take on the qualities we see in Vermeer’s today. Likewise, it’s hard to believe the Vermeer would be satisfied with Tim’s clinical north light transcription.

  17. wwick


    Switching gears a bit, I would like to point out that Vermeer’s ability to capture detail is often sighted as one reason Vermeer is thought have used optical aids. As you well know, there are painters that captured detail far more exacting than Vermeer. However, had Steadman not repeated this mantra in his talk, I may not have brought this up:

    The top photo is a screen grab of Tim at work painting details of the virginal (presumably with the aid of the comparator). The middle photo shows that detail in Vermeer’s original; below that, Vermeer and Catharina’s signature. It would appear that the details of the virginal in Vermeer’s original are not that much more impressive than a signature of an educated man or woman of the day – whether they become painters or not. This may be common knowledge to historians, but in the popular debates over optics in art, the aesthetic milieu in which Vermeer would have been raised, and the kind of skills he would have developed from an early age rarely gets addressed.

    Tim’s movie, and the press that followed, has made much of the fact the Tim is not a painter. My guess is he’s not a calligrapher either. I would suggest the comparisons here hint at why Tim would be far more motivated than Vermeer to find mechanical means to achieve such details. For Vermeer, in his time, it’s more likely such details would be expected.

  18. craig


    I graduated exactly 20 years after you. Hobbs’ class was a real highlight of my school experience.

    I have to say I don’t envy the writing/editing task you face. It sounds monumental. I would be crushed under the weight of it all. From what I’ve seen here, I have no doubt you’ll bring clarity and insight to the subject.

    The experiment you propose makes sense to me, and I think your observation about the black frames (and Cezanne’s hat) is a good one. It will be interesting to see how well painters do ‘eyeballing’ it. In this regard, your experiment reminds me of the project taken on by painter Nicholas C Williams:

    “One area of Dr Stork’s research focuses on the chandelier depicted by Jan van Eyck in the painting ‘Arnolfini and His Wife’ … Hockney and Falco claim the chandelier is in perfect perspective, suggesting this could only have been achieved if traced from an optical projection. ‘To test their assumption, as part of our research’, explains Stork … Williams was asked to paint two complex chandeliers by eye alone.

    “‘Our perspective analysis applied to this painting’ (Chandelier II), continues Stork, ‘resulted in a good but, as expected, imperfect alignment of arms. The average measured deviation was about 8.55 % the image width, of the same order of magnitude as that of Jan van Eyck’s chandelier. This experiment confirms that realistic-looking structures can be painted mainly by eye without the help of optical tools of any sort.’” –from

    However the experiment you propose turns out, we’re still faced with the problem that the claims for accuracy in the Music Lesson – like those for the chandelier in the Arnolfini painting – are not supported by the facts. (Stork summarizes his argument about the chandelier on his ‘diatrope’ webpage). Like you, I don’t know how one could design a measurable, ‘scientific’ test of the comparator idea on a 350 year old painting, but I’d like to think that experts in optics have some ideas on the subject that are worth pursuing.

    I take your point that Tim isn’t an art historian or someone of Hockney’s celebrity, but he did make a film about his experiment that received international acclaim. Also, he made the film with Penn & Teller, whose celebrity and history of science advocacy should count for something. It would be disappointing to learn that nobody in the field of optics has taken them up on their offer of full disclosure.

    If it’s true that no one has done so, I’m happy to be the one to step in and ask. Obviously I’m no expert in optics, but if I had a promise of full access to all Tim’s data, I’m sure I could find qualified people who would be willing to take on the project.

    On the previous thread, Tim gave out an email address. Do you think that is the appropriate place to send the request?

  19. craig


    On the previous thread you made a comment about how, in trying to reconstruct Vermeer’s workshop practice, we’re like South Sea islanders trying to figure out how Wayne Gretzky scored so many goals. “We have excellent athletes. What’s missing is the culture of learning how to play hockey from a young age – the mentors, the coaches, the competition, the venues, the marketplace, the collective wisdom of the community.” Your post about Vermeer’s ability to capture detail seems to me a specific illustration of that point.

    Upon reading it, I was immediately reminded of standing in front of Van Mieris’ “Teasing the Pet” at the Mauritshuis and being astonished at the detail of the floral pattern in the woman’s blue skirt. The rug and rendering of the ermine trim are nothing to sneeze at either. (The Mauritshuis website has an OK hi-res image). I don’t know what specific techniques Van Mieris – and other painters whose ability to capture detail arguably surpasses Vermeer’s – used to accomplish this. It may be part of the unwritten story of painting technique Jonathan was talking about, or maybe I just haven’t done enough research. Either way, your point is well taken.

    Having said that, I do think that using the comparator as a detail transcription device is Tim’s best argument. I confess, this may be in part because I’m like a South Sea islander marveling at how good Canadian kids are at hockey. I certainly wouldn’t mind a mechanical aid in that situation. But more importantly, this is a case where Tim’s claim does seem to match what we see on Vermeer’s canvas. There does appear to be a slight curvature in the seahorse pattern and this curvature does seem to resemble what Tim saw in his comparator setup. The question then then becomes “is this strong evidence that the ‘seahorse smile’ is the result of Vermeer using a similar setup?” Or is it just a coincidence?

    Well, how big a curve are we talking about? If I’ve resized the Music Lesson correctly in photoshop, then the seahorse pattern to the left of the virginal keys is about 5 inches long. With the image aligned so that the top of the virginal is perfectly horizontal, I ran some 1 pixel horizontal lines through the seahorse pattern and checked to see how far the curved lines in the painting deviated from my lines. The greatest deviation I could find was less than a millimeter. Maybe I’m doing this wrong, and I would welcome the chance to review someone else’s probably-more-accurate results, but I’m pretty sure we’re talking about a 1 mm (or less) curve over a 5 inch span.

    If Vermeer had used a straight edge or snap line to make guide lines for his painting, how thick would the lines be? We can’t know for sure, but thanks to the autoradiograph of ‘the Glass of Wine,’ we may have an idea. Claudia Laurenze-Landsberg at the Gemeldagalerie writes: “Interesting lines can be observed in the first and the last autoradiography. These lines become visible because there is no blackening at all. They can only be caused by an underdrawing up to which Vermeer applied paint and left the lines in reserve.” It’s difficult to measure the thickness of the lines between the tiles precisely, but as average peak, the line is about 1mm thick. (I asked).

    So if the deviation from the straight edge line that Tim noticed while painting the seahorse pattern is roughly equal to the thickness of the guide lines Vermeer drew in another painting, should we rush to assume the distortion was caused by a lens and concave mirror?

    Tim said that the distortion in the virginals was, as far as he can tell at this point, unique in art of the period. He had to leave the conversation before he could expand on that point, so I don’t know exactly what he meant. So I went to the National Gallery of London website to take a closer look at Gabriel Metsu’s “Man and a Woman Seated by a Virginal” and was struck by the numerous curved distortions in elements of the painting one would expect to conform to straight edge lines. The virginal in particular. On the same website I found Jan Steen’s “Young Woman Playing a Harpsichord with a Young Man.” To my eye, the top of the seahorse pattern on the harpsichord displays a noticeable ‘frown.’ I could mention several other examples, but I’ll spare you and just say that the curvature in the Music Lesson seems significant when compared to the ruled line on Tim’s panel. When compared to other paintings of the period it appears to me to be perfectly commonplace.

    None of this rules out the possibility that Vermeer did use a lens and concave mirror, but I hope it goes some way toward providing some context to the ‘seahorse smile’ discussion.

  20. Jonathan Janson

    Yes, I believe the general public underestimates the learned skills of 17th-century painters. I can think of nothing more emblematic than that of Gretsky and skilled painters of the past, it is perfect. You should call it “Craig’s Principle” and it should be taught in all art history and art appreciation classes. The problem is that only those that have a sustained in-depth multi-year experience in looking at images and painting or photography can appreciate skill fully. In all my experience, I have found that art historians routinely underestimate, and greatly so, the skills of the past. Works like Van Mieris’ Teasing the Pet, The Brothel Scene are literally, and I mean literally, beyond my comprehension, and not only in detail, but in the delicacy of touch, the play of thin and thick and opaque and transparent paint, sense of contour and tone, and I will stop there. But all this does not mean that Van Mieris had some sort of a mechanical or optical device? It may be slightly narrow to look at only painting for skill. Many decorative artists of the time probably had even more skill that the fine artists. Sometimes even run-of the-mill engravings can be amazing.

    You should definitely contact Tim by email. He’s a busy but an open and energetic man.

    I am not overly-comfortable with the sea-horse-smile concept. Tim once sent compacted images of the Music Lesson and smiles are everywhere, some up and some down, some short and some longer, and some even with two smiles in the width of other broader smiles. I saw no slam-dunk pattern. He did. You could say Vermeer used the comparator here, and not there, but that would be fitting. To be more than coinicidence, it has to be a pattern. The difference, as you point out is a matter of millimeters, perhaps only one, can be caused by various factors. There are, in fact, no straight lines in Vermeer, or at least very few (he did not use a ruler as I see it but by setting the brush handle on mahlstick as a guide). During the painting process, once you have painted the first “straight” line, when you paint those that are immediately nearby and parallel, you will automatically use the first as a guide and the whole series will eco the same curvature or non-curvature of the first, if it is perceptible or not. Because when the two close lines are not “parallel” in their curvature, you can immediately detect the difference between the space between lines (but not the curvature of the lines themselves). This happened to me all the time when I paint buildings with lots of parallel lines, especially with horizontal shingling of US wooden homes. Sooner or later (usually the first actually!) one gets off kilt and all the other follow, the curvature multiplies and becomes noticeable after a while. Then there is the canvas that droops while you apply pressure with the brush–you have to press the brush more as the paint flows out of it to get the same qualitymark–and the canvas that stretches unevenly over time.

    In any case, Vermeer’s painting is not exceptionally detailed, by any stretch of the imagination.

    I was, however, very impressed with the comparator could get the structures of the decorative patterns of the spinet of the Music Lesson so “right” (some of the details are far from perfect) Tim, who had much more practice than I, got them almost perfect, sometime even more accurately that those of Vermeer, although Tim’s are a bit too “graphic.” I went nuts during the first few attempts and was about ready to call it quits but little by little I gradually saw it could be done, and if one really set his mind to get, get them absolutely perfect. But as Tim argues, there is really no sense in getting the abstract patterns of the spinet in the Music Lesson perfect in the first place, if no one would have ever noticed they a were wrong. If Vermeer was able to to so and nobody really cared, he probably found a rather efficient means to do so. But then again, we should look more carefully at the skills of silverware or glassware engravers; they do regular patters with incredible skill.

  21. craig


    As much as I’d like to take credit for the Gretzky comment, that was all Wick. It was such a good analogy, I couldn’t resist the urge to quote his own words back to him. In retrospect, that might have been kinda weird. I think “Wick’s principle” has a nicer ring to it anyway. And I’d be interested to hear if he has anything more to say on the subject.

    I’m all for a historically grounded appreciation of the culture and ‘training regimen’ that would surely have had an influence on painters’ workshop practices. At the risk of sounding like I’m sucking up, I think the work you’ve done on this website and the forthcoming revised edition of your book have done and will do a lot to increase awareness of these factors.

    I sent Tim an email. I’ll let you know how it goes.

  22. Jonathan Janson


    You have discovered one of my greatest talents: overlooking the obvious. Sorry about the name mix up. You know I esteem you intelligence too.

    I have tried to research the “training regimen” as applied to 17th-century painting for my book, but have not found a great deal on it on the web. There is lots about how apprentice contracts but little about the grinding hardships, if they indeed existed. It would be interesting to find something about the other arts too, but I am not too sure where to look. I have access to Jstor, though . I got very interested in the topic reading a piece in Van der Wetering’s touchstone book on Rembrandt’ studio , which I am attaching below. Some could apply to Vermeer. I have often pondered with who Vermeer studied. His early work does not display great command of drapery, foreshortening, anatomy or perspective–on the contrary–so I would wager he did not study with a history painter, or if he did, we wasn’t the most receptive pupil.

    Thanks for the compliments. I am not too sure what contribution I am actually making but I am at least giving it a try. I am almost hoping that the Ebook won’t make many sales so I will have an excuse to upload the whole ?>5%$(!//ed@+*% thing onto the Essential Vermeer free for all, then I could keep on working it like I do on the rest of the site..a perennial state of work in course. I get quite a few visits to the pages about Vermeer’s painting technique so I know it would be doing some good and perhaps stimulate debate. One of my recondite goals has simply been to get on the web everything I would have given my left arm to know when I just walked into Hobb’s class the first time and knew nothing. It took me 40+ years to get some of my ideas straight so I would love to think that some students out there would not have to start from scratch. It’s so ridiculous that with the internet there is not concerted effort to bring things together, like they are doing in some areas of sciences. Both for its relative lack of complexity and volume of information (compared to gene folding or fusion reactors), art history is a very very congenial to being digitalized.

    Anywhere, if you aren’t familiar with it, here is the Van der Wetering thing I was talking about.

    “Painters like Rembrandt were able to grasp the shadows raised by the imagination only by means of a skill that is now scarcely imaginable. If we want to get an idea of the discipline and skill of a painter like Rembrandt today, we would do better not to look at the great majority of our contemporary painters, but at the performing musician or ballet dancer. In these arts, it is still understood that professional skill can only be built up through endless practice from an early age on; and as we saw in Chapter III, the same pertained just as much in Rembrandt’s day to the art of drawing and painting. But it should be added that whoever inquires more than superficially into the careers of those dancers and musicians of our own time who have practised all their lives, will realize that only a few are able to command that unique set of qualities and talents that are necessary in order to develop into major artists. Without that basis of skills, there is, however, no way that this can happen.

    I am convinced that it is the power of imagination and the dream-like spontaneity with which the image is grasped, that provide the key to Rembrandt’s art. There are innumerable paintings, drawings, sculptures, novels, symphonies, ete. that are competently put together, but which are little more than the sum of their parts. With great art, however, it seems that there must have been a powerful image lying at the root of the whole process. The artist appears, as it were, to move with ill that image while he is at work. He is what he is making. The variation in the quality of the handwriting and the differentiation in the rhythm of the hand, the delightful combination of uninhibitedness and appropriateness, seem to be dictated by the force with which that image is kept alive in the artist’s imagination. ln Rembrandt’s case, as discussed in Chapter IV, there are indications that he did in fact conceive his works from imagination. And whether he subsequently worked from nature or gave form to a conception born purely of fantasy would seem to make little difference after that. “

  23. wwick

    Craig and Jonathan,

    I think we could call the analogy “Gretzky’s Edge”. No worries about the mix up, I was just happy to see it resurrected since I assumed it fell on deaf ears when I originally posted it. Jonathan, I’m also happy to see that Van der Wetering’s analogy amounts to somewhat of an affirmation. I chose a sports analogy because Tim’s Vermeer (the film), like Hockney’s Secret Knowledge, seems to be gunning for a general audience who would be familiar with endless hyperbolic praise lavished on sports heroes, and I thought it disingenuous of the film’s producers to frame Vermeer’s accomplishments as literally supernatural, as if to suggest his skill was beyond the realm of mortals. Truth be told, I know little about sports and less about hockey, but as it happens, Gretzky, The Great One, the yet to be surpassed highest scorer in NFL history, was described as a “magician” in part because he was otherwise lacking in size, strength and speed.

    On the matter of the virginal’s calligraphic detail, I would suggest that if indeed, educated men and women were expected to master calligraphic writing from an early age, why would Vermeer, having acquired that muscle memory, then want to master a new, complicated set-up to accomplish essentially the same task? Once the training wheels come off you just want go. Then there’s what I call the chicken and the egg problem. How did the decorative artist achieve the seahorse design in the first place? In Secret Knowledge, Hockney is astounded by Holbein’s detailed rendering of a painted globe, but leaves unexplored the problem of how the detail on the original globe itself came to be so intricately rendered. Thanks to Tim’s experiment, we now know the resolution of a primitive camera obscura is not sufficient to render such fine detail – pretty much demolishing Hockney’s theory on that score. The comparator is the ad hoc solution that saves optics, if not the camera. But increasingly, I’ve come to think the only necessity for this device is to satisfy a present day culture who can’t imagine any other solution.

  24. craig

    Jonathan and Wick,

    “Gretzky’s Edge” FTW.

    Reconstructing what the ‘training regimen’ for painters might have been is a real challenge. It seems reasonable to assume that different painter’s workshops emphasized different skills. This may explain why we sometimes hear of young apprentices studying with several different painters. Nicholas Berchem, for example, is said to have received instruction from six, including his own father. But here also, it’s probably best to avoid making too many assumptions about what particular skills painters passed along to their students. If we had only the paintings to go by, would any of us guess that de Hooch learned his trade from Berchem?

    Just focusing on the apprenticeship system misses a lot. I think you both make a good point in calling attention to other, related fields which could plausibly have had an influence on a painter’s work. I’m reminded of a few lines from Jorgen Wadum’s essay in the 1995 Vermeer exhibition catalogue:

    “In the introduction of his book on perspective, Desargues writes that a painter who wants to know more about the Meet-konst (art of measurement) should consult the Landmeeter (cartographer) in order to make use of his expertise. According to Desargues this would lead to a better understanding of Doorzicht-kunde (perspective). He further suggests that the painter should look around him at other crafts and take advantage of the knowledge of for instance, carpenters, brick layers, and cabinet makers.”

    I don’t know if there’s any documentation explicitly describing a sharing of skills between the various trades. Maybe it was all just informal, so we’re left making educated guesses. It’s widely accepted that Saenredam learned perspective after he completed his 10 year (!) apprenticeship. Cartographer Pieter Wils generally gets the credit, but I’m pretty sure that even this is just informed conjecture.

    Having more primary source material freely available online would be a big help. I’d also love to see more technical information (pigment analysis, x-rays etc.). Jonathan, the work you’re doing on this website (and your book) goes a long way toward meeting that need, but really, wouldn’t it be awesome to have unfettered access to the full technical dossier on all those paintings? Back in the day, I was looking for that stuff because I figured it would help me do a better job making Vermeer copies to give as wedding gifts to friends and family. Selfishly, I still want that access, but the reason for it has changed.

    I hope you don’t mind, but I’d like to explain this shift. It may be a far-too-specific way of agreeing with your point about wanting more art historical documents to be digitized, but in the end, it may be of some use to the larger conversation. Either way, I should apologize in advance, because it will take a few posts to get it all out.

  25. craig

    Earlier in the thread, I mentioned the confusion I always felt over the anomalous sizes of the windows in several of Steadman’s reconstructions. The ‘Wine’ paintings in particular always struck me as problematic. To put it another way, I thought Steadman was spot on when he wrote “we are stepping on to dangerous ground here” by suggesting that in both paintings the finished work was a collage of two projections. This is not to suggest that Steadman hasn’t earned the respect he’s rightly received for his work, but frankly, that aspect of it always bugged me.

    Since Tim’s project relies so heavily on Steadman’s work, the buzz surrounding the film got me thinking about it again. I knew the Glass of Wine had a pinhole marking the vanishing point, but I’d never seen a reproduction that indicated its precise location, so I emailed the Gemaldegalerie in Berlin to ask if they had an image that could help me out. To my great surprise and delight they had a new x-ray taken and sent me the image. Since I’m no expert in reading x-rays, I double checked with them to make sure I had the pinhole location right. Assured that I did, I then worked up a perspective analysis on a photograph of the painting that they also provided.
    I’ll post the image of it below, but since it includes distance points, the image is quite long and may be tough to read at this size. Anyway here goes…

  26. craig

    There’s a lot of stuff going on in that image, and it’s hard to resist the urge to talk about it all in detail, but for clarity’s sake, I’ll focus on the point where the horizon line meets the right edge of the painting. As you can see, this point seems like a reasonable vanishing point for the open window. Interestingly, it also marks half the distance between the principal vanishing point and the distance point to the right of the canvas. If Vermeer constructed the perspective in this painting rather than trace it from a projection, this would be an ideal spot to find a construction mark.

    I knew that the Kunsthistoriches Museum had recently discovered a pinhole on the horizon line at the right edge of the Art of Painting, so I thought maybe there’s a pinhole on the right edge of this painting as well. I emailed the Gemaldegalerie again to ask if there was indeed a second pinhole. Amazingly, they took another photo, which I’m posting below. As you can see there’s no pinhole, but there is a distinct mark in that spot (circled in red) indicating that something flat and circular had been pressed into the primed canvas.

  27. craig

    In this last image, I’ve traced the round mark as well as the outlines of the figures in yellow and superimposed them over my original perspective construction.

  28. craig

    I’ll admit, it was pretty cool to see a mark on the canvas where I thought we’d find one. But then again, I was expecting a pinhole and this was something different. I don’t know what caused this round mark. It could be that something random was pressed into the canvas and its location is a matter of pure coincidence, but I’m tempted to think it isn’t. If we assume that this mark beneath the paint layer was left by Vermeer as he constructed the perspective, then the layout of this scene starts look a lot like what you see in old perspective manuals. Even the way the corners of the tiles line up along the bottom edge of the painting looks like what we see in the Hondius print here:

    Still, at this point that round mark is just a physical fact without enough context to assure me that I understand its purpose. Is the mark unique to this painting, or are similar marks lurking unnoticed beneath the paint layer of other paintings? If it is a construction mark, then this raises the possibility that Vermeer used a simple method to lay out the floor tiles that would not require use of distance points outside the piece. Was this method used by other Dutch painters? If so, there may be a physical record of it on their paintings as well.

    This, at last, brings me back to your point about the benefits of digital access to art historical resources. If there’s a study that’s been done on the practical methods of perspective construction among members of the Delft School, I haven’t been able to find it. I assume individual museums have accumulated a wealth of information on particular paintings, but I’m not sure how someone like me can gain access to it. I was presumptuous enough to email Jorgen Wadum to ask his advice and assistance on this (twice), but never heard back.

    That experience made me realize just how spoiled I was by the Gemaldegalerie. When I first emailed them, I really didn’t expect a response. Before I knew it, they were taking x-ray photographs of a Vermeer painting to answer my pesky questions, which is frankly astonishing. I don’t think I can do enough to thank Claudia Laurenze-Landsberg for her kindness, patience and willingness to engage on this subject. She could easily have said “look, buddy, I just sent you a 20meg TIFF of the autoradiograph. You should be happy with that.” Instead she had the new x-ray taken and found the round mark. If anything does come of this, she certainly deserves the credit.

    For now, I admit, I’m feeling stuck. I may just start emailing other museums and hope that I’m as lucky ‘cold calling’ as I was the first time. If you can think of another approach, I would certainly welcome the advice.

    I’m sorry to have gone on for so long on this tangent, but I hope you can see why your comment about the need for digital access to art historical resources struck a chord with me.

  29. wwick

    Craig, this is excellent work. The mysterious dimple is certainly suggestive. Equally important, however, is the elephant in the room you brought up: the disparity of window sizes between those of the two “Wine” canvases with that of the “The Music Lesson”. Your perspective reconstruction reveals something else too: it’s unlikely Vermeer collaged two projections together, as Steadman suggests. Here’s why:

    According to the professor’s calculations, the projection for “The Wine Glass”, when focused on the center of interest, would coincide with the back wall of the studio. Assuming the center of interest is the table, chairs and figures, upon tracing those elements, Vermeer would then have to move the lens back to make the window fit within his composition. In doing so, everything would get smaller. But he would then have to move the canvas backward to refocus the window, which he can’t do, because the back wall prevents that. Furthermore, it should be noted that any existing tracing would become out of register with the smaller projection, causing visual confusion in evaluating the resized window together with the first tracing. Even so, with the vantage point now further back, the perspective lines of the window frame and sill would no longer converge at the same vanishing point of the rest of the sketch, as it clearly does here.

    While I appreciate that professor Steadman appears to have been scrupulous in articulating inconstancies where he found them, his summaries lack the same rigor:

    “It is important…to re-emphasize the extreme sophistication and precision of the underlying perspective geometry of Vermeer’s pictures” (Vermeer’s Camera, page 154)

    Given that the “Wine” paintings have a non-conforming window size, the Music Lesson has a ceiling height some 20” taller than two other paintings, the size of the virginals are all over the map, and the difficulty of verifying heights of tables and chairs without the having actual items in hand, its hard to justify the implication of “extreme” precision (close agreement with some maps and globes notwithstanding).

    He goes on:

    “If Vermeer worked mathematically, why would the ’projected images’ on the back wall of the room work out to the exact sizes of the paintings themselves?” (page 155)

    Once again, given the internal inconsistencies within those projections, it’s difficult to understand how the word “exact” is justified. Nonetheless, he is so strongly tethered to the back wall theory, it’s difficult for him to entertain anything else. Earlier in the book he offers a number of reasonable solutions for a correcting the CO projection’s left-to-right inversion problem by using mirrors to re-orient the image on a horizontal surface. He then writes:

    “The main objection for this present argument is that any geometrical reason for the coincidence for the six ‘projected images’ on the back wall then disappears” (page 113).

    Now that Tim has entered the picture, the back wall theory itself is on dangerous ground, because to what extent the back wall projection is needed for tracing is not entirely clear, as Tim did not use it, as far as I understand. Steadman envisions the camera obscura as a “composition machine”. As one who has spent years under a dark cloth of a large format camera, I take issue with the hubris of this notion. The awkwardness of collaging two projections illustrates the problem perfectly well. There are many other reasons too, but I will leave it at that for now.

    Thanks so much for this post, Craig. I encourage you to continue, and if you do so with the same clear-headed approach you’ve shown in this thread, many more museum doors will surely open for you.

  30. craig


    Nailed it again with your coining of new terms. “The mysterious dimple” captures the essence of the thing perfectly. If the blog had the edit function you asked about a while ago, I’d be tempted to go back and replace the rather dull sounding “round mark” with this better descriptor. Though really, I might just be thinking about editing because I wish I could revise another sentence. The one that begins “As you can see, there’s no pinhole…” should have read “There’s no clear evidence of a pinhole, but….”

    The dark spot on the x-ray just above and to the left of the mysterious dimple is indeed a hole in the canvas. I checked and it does seem to provide a more snug fit for some of the lines on the window, but it resembles other (random) holes in the canvas. It’s larger than and not as round as the pinhole at the principal vanishing point, but I’ve been told we can’t rule out the possibility that it was a pinhole that expanded when the canvas was mounted on the stretcher.

    The mysterious dimple does provide a reasonable vanishing point for the window, but it’s not a slam dunk. The dark spot/hole (“peculiar perforation?”) underscores how ambiguous the evidence is and why I’m so keen to get more context to help me understand what we’ve got. My latest email request for advice and assistance seems to have failed as well. I don’t know if I’m not getting past the spam filters or if the people I’m writing to think this is all unworthy of comment. Either way, I’ll keep at it…

  31. craig

    In the meantime, I agree that it’s a good idea to talk about the elephant in the room. When you wrote that “given the internal inconsistencies within those projections, it’s difficult to understand how the word ‘exact’ is justified” you gave voice to exactly what I was thinking when I finished ‘Vermeer’s Camera.’

    I assume Steadman is tethered to the back wall projection idea because that’s what makes his theory compelling. It’s arguably the essence of his theory. Steadman mentions how Paul Taylor and Tim Gowers tried to calculate the probability of the six paintings being projected at their actual sizes on the same plane and came up with odds that are “hundreds or even thousands-to-one against.” I have no idea how one calculates such a thing or if their calculations took the inconsistencies into account, but their conclusion seems to confirm a common sense reaction to the coincidence. I get that. Still, I’m disinclined to just wave off the inconsistencies.

    Steadman’s starting assumption in working up his calculations was that the floor tiles in all these paintings were exactly the same size (with the ceramic tiles being half the size of the marble tiles). I hope someone will correct me if I’m wrong about this, but as I understand it, working his calculations this way is what gives him the amazing coincidence of the six paintings being projected at their true size on the back wall. If he were to use the lions head chair from the Prisenhof Museum as the key in determining the relative scale of objects and the room, would the projected images still line up on the same plane? Looking at Steadman’s measurements for the chairs in the Music Lesson, the Glass of Wine and Lady Writing a Letter with her Maid, it’s hard to imagine how they could. I hope someone with a real understanding of the math involved could help clear this up.

    What struck me as strange about Steadman’s method was that he used the thing least likely to have actually existed in the room as the key in determining his measurements. On page 115 he writes “The possibility emerges that the real floor of the room was the ceramic tiled surface.” But how real is this possibility? Vermeer’s studio was above ground floor, and my understanding is that tile floors were generally restricted to the ground floor. How many examples of above-ground-level tile floors in 17th century residences exist in the archival and archaeological record? Also, did Dutch ceramic tiles tend to have a side length equal to exactly one quarter a Florentine braccia? That seems like a pretty amazing coincidence.

    These questions and doubts have always bothered me in part because I’ve never seen them directly addressed. This could be because I never looked in the right places, but I hope someone can address them here. You’re right to point out that Tim’s experiment raises a whole new set of questions about the back wall theory. Tim’s mirror method does seem to make the geometrical reason for the coincidence for the six ‘projected images’ on the back wall disappear. Also, his experience tracing the tiles raises serious questions about the wisdom of using the tiles as a key to working out the geometry.

    There seems to be real dissonance between Tim and Steadman on this crucial question regarding Vermeer’s use of optics. I would very much like to see it openly discussed. A public discussion between Tim and Steadman on this narrow topic would be very informative.

  32. wwick


    To be clear, I don’t dispute that there’s enough geometric conformity among 10 or 11 paintings to suspect there’s measurement going on, I just don’t see the back wall theory as definitive. It’s merely suggestive (like the mysterious dimple), and holding on to it so dearly, as Steadman clearly does, is a dead end. We can always come back to it; the assertion need not go away completely. I just think it’s healthier to point out its weaknesses, and continue to look at other solutions. In the meantime, I would think if the geometry is as exact or precise he claims, it could be confirmed with 3D models made with Lightwave or Sketchup. Yes, we have the 1/6th scale models Steadman made – it showed the “dangerous ground” of the non conforming window sizes – but other differences are commented on as well. It’s just that it’s hard to SEE those differences in verbal descriptives or data tables. Visualizing the 11 paintings with 3D computer models, and overlaying the wireframes to a single master (such as the Music Lesson), one at a time, would help a lot, even if much of the furnishings were left out. In theory, only ONE wire frame model needs to be built, since it’s supposed to be the same room.

    I agree the tiles are the least likely features to have been present in the studio. However, is it possible that he might have faux painted his studio floor? Not for every different patterns we see, but a single checkerboard pattern that would be useful as a built in guide for “placing” furniture on his canvas. Another possibility is that he had Delft tile skirting at the base of the wall. Even without a floor pattern, they’d make a great reference for visual measuring.

  33. craig


    I think we’re on the same page. Your proposal for 3D modelling makes sense to me. Not only would it be useful to really see what we’re working with, but I imagine it would also allow us to ‘game out’ various possibilities. I’m not familiar with the specifics of each modelling program, but I’m guessing that once the master template is complete it would be fairly simple to make adjustments to see what happens when we assume all the chairs are the exact size of the Prisenhof chair, or if we make all the windows the same size. Seeing all of these possibilities play out may suggest answers to some of the questions that have been raised. So yes, it’s definitely worth doing. If only we knew someone who was familiar with Lightwave and had an interest in this topic….

    I agree that there’s benefit in setting aside the geometrical coincidence of the 6 paintings while evaluating all the evidence, but there’s no denying it will always be the elephant in the room. It’s what Steadman points to when dismissing the idea that Vermeer used geometric construction methods. Though he seems to be dismissive of the idea in general. On page 154 he writes “The dimensions of the room and of recurring items of furniture, as we have seen, are represented consistently from one painting to another. In order to achieve such results by following mathematical methods, Vermeer would have needed to prepare accurately measured working drawings both of the room itself and of all the items depicted in it. A geometrical technique is conceivable, but it would have been far from simple.”

    I confess, I’m not sure why I should be surprised or amazed that a competent painter is capable of representing chairs chair-size and window sills at sill-height. Is it really so hard to achieve such results by eyeballing it? If Vermeer did use a geometrical technique, wouldn’t it have been a lot simpler to work from measurements of a corner of one’s own studio than say a church like St. Bavos? The task Saenredam posed for himself was vastly more complex than Vermeer’s simple, hermetic scenes, and yet somehow he managed. And it seems other painters of church and domestic interiors did so as well. If it really was so much easier to just trace a camera image of an interior scene, why was Vermeer the only one to do it?

    As part of his attempt to dismiss the possibility that Vermeer used a mathematical method of construction, Steadman calls attention to the fact that Vermeer’s vanishing points tend not to coincide with meaningful points in the composition. “With a camera, by contrast,” he writes “the position of the vanishing point tends to fall more fortuitously (although a camera can be lined up with care to centre on some chosen point, if required).” It’s a compelling argument, until you start locating the vanishing points in the paintings of Vermeer’s contemporaries. It turns out fortuitous vanishing points were the norm among Dutch genre painters. So, on this point, Vermeer does not stand out.

    Steadman also discusses the pinholes that mark the vanishing points: “I have proposed that Vermeer might have made tracings on paper with the camera, and then transferred these to canvas. A pin at the central vanishing point would have been as useful here as it would for perspective drawings set out mathematically. I have found it helpful myself, when making perspective analyses on tracing paper over reproductions of paintings, to stick pins at the vanishing points.” Tim, for slightly different reasons also found the pin at the vanishing point to be helpful. They’re both right. The pinholes can be useful for traced drawings. But let’s be clear about what this argument implies, which is that de Hooch, Houckgeest, and all the rest used pinholes for their geometric constructions, but Vermeer used his on a traced image. That sounds an awful lot like a special pleading to me.

    So maybe it’s not such a good idea to look at Vermeer in isolation. Far more useful, I think, would be an approach which investigates the perspective methods of the Delft School as a whole, including the architectural painters. If Vermeer’s approach really was different, then a full review of the available evidence will make that clear.

  34. wwick


    I agree we’re on the same page (perhaps the few in the world who are!).

    Not to belabor this (or that you would need any convincing), but for the record, I’d like to further articulate problems with the issue of composing and refocussing using a booth-size camera obscura. To do so, I’ve appropriated a photo Tim took (a detail of his Music Lesson projection) to simulate what this would be like in practice. I’m showing the projection upside down AND backwards, because that’s how Vermeer would have seen his projection inside his booth-size camera (a difficulty Steadman has acknowledged).

    The top picture simulates a little ambient light coming into the camera. It would begin to wash out the colors, but would also make it easier to see the tracing over the projected image.

    In the bottom photo, I’ve simulated more ambient light coming into the camera as one might do in practice to see the drawing more prominently. I’ve “refocussed” the projection, reducing it by 6%. The drawing is therefore misaligned, which creates a certain amount of confusion in evaluating the merits of the new elements to be traced. For example: imagine erasing the traced windows, then retracing their new position. You can see how the influence of the projection would throw off your judgement. I’ve had a lot of experience with this, because in the pre-digital era, there were many reasons I needed to trace from projections on the back of an 8×10 camera. It’s not impossible, but it’s far from an ideal way to work.

    Composing a picture implies experimenting with different arrangements of major and minor elements of a scene. At this stage of the process, any one change begets another, and so on, and it can be very frustrating to compose by moving physical objects on a set while struggling to see in the dim light of a projection. Vermeer could pre-arrange much of it first, but once inside the booth, he’s powerless do do anything but look at the screen and make verbal instructions to assistants. He would then have to wait as they come into the scene, make the changes, and step out again before he could evaluate it. With each new change, another object might need to move, entailing more waiting. In practice, it’s excruciating to work this way. Ideally, you want make the changes as fast as you can think of them. To add to the awkwardness, Vermeer must evaluate his projection obliquely, because his head would block the lens if he looked straight on to his canvas, and he could not step back from the image more than 2 feet, because the booth is only about 3 feet deep.

    Certain draped elements, such as Vermeer’s artfully styled table coverings and costumes, cannot easily be styled by way of verbal instructions. It’s true, he could step outside of the camera and style them to his liking, but he has to crawl back inside and wait for his eyes to adjust. And with the projection being upside down, the draperies would be disconcertingly “falling up” (and backwards). Furthermore, when he needs his models to step into the scene, they would have to respond to a disembodied voice coming from inside the camera rather than get the benefit of the artist’s eye contact and hand gestures to guide their poses. Add to all that, the focus restrictions imposed by the back wall, the extreme softness due to shallow depth of field and primitive optics, plus the extreme darkness of the projection.

    Steadman has imagined for Vermeer, a working process that’s awkward in the extreme; a process for which there would have been no known precedent – no tricks or short cuts passed down from one artist to the next – as would be the case with generations of perspectivist (the best of whom can be found in Delft when Vermeer was at an impressionable age). He would have had to devise this camera system out of whole cloth, from the ground up. The supposed benefits of the flattening effects of the projection, or the claim that the inverted image would help Vermeer to see his composition abstractly in terms of pure shape, are nothing more than art school tropes, in my opinion, and would be of little benefit in actual practice. And worst of all, that Vermeer’s art, would have been influenced by this process – the stillness of his subjects, the lack of obvious line, the fall off of light – is the tail wagging the dog. If only it were that easy: to pick the right equipment.

    In light of the these many problems, it’s worth having another look at Steadman inside the booth camera Tim built, and imagine in Vermeer in his place. You’ll find the still at 30:58 into the lunch hour talk:

    Now, It would behove me to suggest how Vermeer did work out his compositions and end up with paintings with the geometric conformity he apparently has achieved. Like you, I’ve been thinking how he might have adapted techniques used by Delft perspectivists. I’ll lay out some ideas in another post to see what you think. I’m hoping you might have something to add to this from a painter’s point of view.

  35. craig


    As you know, cameras and lenses are outside my field of study, so it’s really helpful to hear someone with your experience talk about the practical issues involved in the working method Steadman proposed. I suspect that Steadman and Tim would like to raise a few counter arguments. For someone like me it would be useful to hear some kind of round table discussion so all this stuff can get hashed out. It would also be useful to have an expert in optics moderate the discussion to resolve any basic questions of fact.

    For example, on page 141 Steadman writes “The most serious of the image distortions produced by a simple uncorrected lens occur, as Mills shows, around the edges. But the effects of vignetting can to an extent be overcome in the camera obscura by refocusing so that the periphery is in focus (and the centre not) – although brightness is still lost toward the edges. This is achieved by moving the lens away from or towards the projection screen. A further possibility is for the lens to be tilted (while the screen remains fixed). This has no effect on the perspective, but can bring a band right across the entire image into sharp focus, following what is known to photographers as the Scheimpflug rule.”

    Liedke took issue with the claim about tilting the lens, writing: “The actual effect of decentralizing the vanishing point in a camera obscura is dramatic, as explained and diagrammed by Kitao.” It seems to me that Liedke and Steadman can’t both be right about this. One of them must have made their argument based on a misunderstanding. But which one? The only thing I know for sure is that I’m not qualified to give an answer. I can’t help but think an independent expert should be able to resolve this dispute to everyones satisfaction. It shouldn’t be hard to form a consensus about facts.

    Of course, the quote from page 141 also discusses refocusing. If the image you posted is indeed representative of what happens when refocusing a camera obscura lens, then it would seem that the attempt to overcome one image distortion simply creates another. Again it would be helpful to hear Steadman and/or an independent expert respond to all of this.

    I’m looking forward to what you have to say about how Vermeer may have adapted techniques used by Delft perspectivists.

Leave a Reply