To Do: Migrate

This lecture was presented at the 3D Digital Documentation Summit held July 10-12, 2012 at the Presidio, San Francisco, CA

3D Scanning of Matisse, the Back I-IV: One Thing After Another

Matisse created Back, his largest sculpture in 1908/1909. Matisse resculpted Back three times over the next twenty-one years (1913, 1917, 1930/31), each time beginning with a plaster cast of the previous state. He also preserved a plaster version of each state, and, cast in bronze they are the works we know today—Back I, Back II, Back III, and Back IV.  The bronzes were cast beginning around 1948 and the edition was completed over 30 years later, around 1981.  Each of the four versions of Back is an edition of 12, making the total number of casts 48.

Until 1954, only three versions were known: Back I, III and IV.  After Matisse’s death in 1954, a hitherto unknown Back, today known as Back II, was discovered in his warehouse, ready for shipping to the foundry, but never sent. Based on descriptions from the Matisse family and formal art historical analysis, art historians had always assumed that the reliefs were made sequentially, basing the next version on a plaster of the previous version. However, immediately after the surprise discovery of Back II, even Matisse scholars were unsure of its placement in the series.

The goal of the recent research into Matisse, Back I – IV was to determine the sequencing of the Backs.  The examination was performed on the four MoMA bronze casts. Laser scanning and close visual examination of minute casting details proved that the casts were sequential and Matisse left large parts of the background alone as he reworked the central figure. In the background, features that appeared on Back I were repeated on Back II. Features that appeared for the first time on Back II were repeated on Back III. Features on Back III appeared on Back IV and nowhere else.

The 3D data captured from laser scanning empowered researchers to analyze and compare the sculptures using CAD software.  Each sculpture was scanned independently, with particular attention to the selected features that could confirm the sequence. Digital models of each sculpture were aligned in a common coordinate system by matching up the selected features.  Once all four sculptures were aligned, overlays, color maps, cross sections, and animations provided the insights into the relationships of the sculptures.

An overlay is an observation of two or more of the sculptures occupying the same virtual space simultaneously. By adjusting colors and transparency levels, similarities and differences become readily evident on a qualitative levelA color map is a dimensional deviation plot which presents the sculpture colored with a spectrum where the colors represent the deviation between two sculptures.  This is a more quantitative method of comparing two sculptures. The most telling comparisons came in the cross sections. By virtually slicing the aligned digital models through x/z and y/z planes into cross section lines, similarities and differences in each pair of sculptures were illustrated.  Figural changes were major but the backgrounds were often identical from one sculpture to the next. The data obtained by laser scanning answered the questions of sequencing, and the method of artistic creation.


3 – D Documentation – Lynda Zycherman

Church:    This is 3-D scanning of Matisse Back 1 through 4, One Thing After Another, and today we have Lynda Zycherman and Joe Nicar. Lynda studied at the Conservation Center at the Metropolitan Museum. Lynda joined the Sculpture Conservation Laboratory of the Museum of Modern Art in 1984. She has researched a wide variety of topics including minimalist, sculpture utilizing electric lights, fluxes, and Brancusi’s bronze sculptures.

Joe graduated from Beloit College in 1996 with a Bachelors of Arts Degree in Anthropology. Before joining Direct Dimensions earlier this year Joe worked all around the south US on archeology digs for the Navajo Nation working on ruins stabilization crew for the National Park Service.

Zycherman:    Thank you Jason.

Henri Matisse sculpted these bas reliefs Back 1 through Back 4 from 1908 through 1931, a span of more than two decades. In them, we can see him develop some of the formal aspects of modern sculpture as well as his own personal vision. He moves from something relatively naturalistic in the first to volume changes slightly less naturalistic in the second to geometric abstraction and finally at the end, what we would all recognize as full abstraction.

Today these are considered among the most important monuments of modern sculpture. However they were practically invisible until the 50’s. Only Back 1 was exhibited as a plaster shortly after it was finished in 1912 and 1913. Four decades later in 1948, Matisse was rich enough to have a bronze cast of number one and he showed it at an exhibition.  After that, nothing until the 50’s.

This photo taken in MoMA’s Sculpture garden gives you an idea of the scale of these massive reliefs. But even better is this shot. This is Sotheby’s Chairman, David Norman, somewhat over enthusiastically discussing Back 1, part of a set that was sold at auction last year. Each back is over six feet high and a little wider than four feet and they weigh at least , the MoMA’s set, anywhere between 300 and 800 pounds depending on the foundry and the alloy.

In preparation for an exhibition and catalogue in cooperation between the Museum of Modern Art and the Art Institute of Chicago, the art historians and conservators wanted to know as much as possible about the making of the Back. Technical examination and close examination were brought to bear at the MoMA lab. Some of the questions we had were;
•    In what material were the Backs originally sculpted?
•    Did Matisse do it?
•    What were his working methods?
•    Are the backs physically sequential?

The reason that’s important is because Back 2 was not discovered until 1954, after Matisse’s death. When it was first brought out as a plaster, the art historians actually weren’t sure where its placement was in the sequence.

The story of the Backs begins here in Matisse’s studio and sculpture school in Sacre Coeur, Couvent du Sacre Coeur. This photo was taken around 1909. Matisse is on the right, the fellow with the beard and the long smock. All the items for sculpting and clay are present, which is the sculptors stand and beneath it a bowl containg water and a cloth which was draped over the sculpture to keep it moist when it wasn’t being worked on. This photo shows small scale items but Matisse had something much, much larger in mind, an over life size bas relief.

He began this Back, the largest sculpture he ever made in spring 1908 in that studio in Paris that I showed you and he worked on it for over a year until June 1909. He prepared to move to a new studio in the suburbs of Paris and he had the clay relief cast in plaster because he was afraid that it wouldn’t survive the move. Making a plaster cast of clay destroys the clay and so this photo is the only extant evidence of the first iteration of the Back, now usually called Back Zero.

Matisse’s working method was typical for the creation of large clay reliefs. On large wooden boards, he hammered large headed nails and if you look on the left side and you see small circles, that’s what those are. He rolled out sheets of clay and pressed them on the board and then created the entire image. Those nails are not only evident in this photo of the clay but even later on in the plasters that were made from it and the bronze cast that came after that. So this sort of makes the sequence clear. Back Zero in clay on the left. To preserve it when he moved his studio he had it cast in plaster, which destroyed the original clay.

Since we have plasters and bronzes of Backs 1 through 4, why is there no plaster of Back Zero and a subsequent bronze cast? The answer is because Matisse worked directly on that plaster of Back Zero and irrevocably transformed it into Back 1, which is what you see on the right side. My opinion is that he regretted not having a record of Back Zero and from now on, he had two casts made when he decided he was finished with a state. One he kept in the warehouse and one he kept to work on further. Four decades later that plaster of Back 1 was turned into the bronze and his work pattern is set.

Matisse worked on the reliefs in plaster and not clay. He added fresh plaster to the surface of the cast and then sculpted it with chisels, hammers, rasps, and so on. The stored plaster version of each previous state is what we have and that’s why we have the four next iterations.

A few of the background features from clay Back Zero are actually preserved through the reworked plaster of Back 1 and subsequently into the bronze which is what you see on the screen on the right. The ones you can most easily see and I think the room is not quite dark enough; the lower left hand corner is the signature Henri Matisse. On both of those, you can see the nail heads, representing the nails that were hammered into the wooden support and there is this odd little rectangular plate which is also a persistent feature.

Besides features that were easy to see in close observation, we wanted more and so we thought 3-D scanning was the way to get it. We wanted a detailed view of the surface features, the ability to superimpose one Back over the other to observe changes in the form and measurements of the relative depths of the sculpture. So now we get into the more technical thing and I’m going to hand it over to Joe.

Joe:    Okay, so in this slide here, this is Glen Woodburn from Direct Dimension. So he spraying the sculptures which are in bronze with developer, which is essentially just talc suspended in alcohol. We decided that it was going to be okay to do that to these sculptures because they are displayed outside, so they already have been worn a little bit. We have come up with other methods that are a little less chemical but given what we wanted for resolution, we decided that it would be okay to use developer on these.

Zycherman:    Let me just say that they are very, very heavily waxed because they are outside so I knew that I could get this off and indeed it all came off perfectly well.

Joe:    Here are the sculptures all sprayed prior to scanning. For this project we chose to use a
Faro arm with an LDI scanner attachment. You can see the setup there on the right with the wheel dolly. For those of you who are unfamiliar with the scanning arm, once it’s calibrated the arm creates virtual space, a virtual coordinate system. Each elbow of the arm you see there has an optical encoder that relays the position of the probe at the end of the arm or in this case when you attach the laser scanner, it relays the position of that laser scanner, so your laser scanner is being operated from the end of this arm so it greatly increases the range, allows you to move around an object. The accuracy, you can see here, is listed as .057 millimeters under ideal conditions obviously, which none of us ever work in.

The onsite scanning took about two days. So we captured the scans at half millimeter resolution over the course of two days. Here the scans are registered into a local system. This was done on site for all of the Backs to make sure we got all of the data we needed to complete the project. So once all the scans were taken, they were taken back to our office and all of them were meshed to a water tight model. Here you can see all four of them. There all meshed at sub-millimeter resolution and at this point they were ready to be inspected.

Zycherman:    So the first feature is a detailed view of the surface features. Here again is Back 1 in bronze and it’s artificially colored and artificially lit grey scale and Back 2 and I think you can all see that it’s actually much easier to see the figure against the background in these artificially created scans than it is in person. They’re very black, they’re very shiny and they’re very outdoors. When you compare the two scans, you can magnify areas to study the surface more closely and where the red arrows are points out details that are preserved between Back 1 and Back 2. Keep in mind that when I’m looking at those bronzes and the scans of those bronzes, I’m actually mentally looking at the plaster casts that Matisse sculpted, not the cast bronze. I mean that’s what I’m looking at but I’m looking behind it to see the plaster that he made.

The molds and scans that Matisse used to preserve the current version and give him the second version to work on, transferred and preserved very small details from the previous version. So two areas in this comparison are in the curve of the left wrist you see the parallel striations of the finger marks. They are identical in Back 1 and Back 2. And the shape and details of the hand and thumb are also identical. So you can see how skilled the mold makers and bronze casters actually are in that the molds made from the plasters that Matisse actually sculpted picked up the incredible number of details that went through two iterations and come down to us as the bronzes.

To further prove the sequencing of the Backs from previous states of the plaster reliefs, we can also look at the upper right hand corner of Backs 1, 2, and 3. So in Back 1 and 2, which I pointed out with the blue arrows, you can see the same V-shaped hollow. New to Back 2 are the chiseled diagonal lines in the upper right hand corner which repeat only in Back 3 and are gone by Back 4. Note the tiny casting details that also appear in Back 2 and 3.

Crenulations, pits, raised areas, these are the result of molding and casting and cannot be duplicated by carving. Secondly, the ability to superimpose one back over the other to see changes in the form.

Joe:    The matching details that Lynda was pointing out to you there, in addition to all of these here on the list; they came in handy for the registration as well. This list was used to point us in the right direction of matching areas between each version of Back. We’ll point out a few of them right here. So what we did was, using the registration software, we registered one to two, two to three, and three to four. Using these areas that we believed had congruent between the two versions and then ignoring all of the other data, anything that was unique to the individual Back was excluded for our registration purposes. So there you see the signature in Back 1 and 2 and then between Back 2 and 3, we have those striations that she was pointing out on the upper right corner. Here between three and four, the right plane, as you are looking at it, of the hand remain the same between three and four.

Zycherman:    Actually the only detail between three and four that are the same is this small
area on number four. Once all the versions were aligned in the same coordinate system, we were able to track the progression of form and the metric geometry of change. Each back is assigned a color as you may have guessed already, blue is one, two is yellow, three is red, four is purple. But I’m going to concentrate on one and two in the interest of time.

We selected a few horizontal and vertical cross section lines for direct comparison. The line I’m looking at is marked by the red arrow, it’s eight centimeters from the left edge and it the feature I’m looking at is the background only. You can see that the background contour is nearly identical in number one and number two, the blue and the yellow. There is a small difference at the bottom ledge which could be the result of a casting deformation or post casting handwork.

If you move on to the figure, again the red arrow on the small image in the upper left, shows you where I’m at and what I’m looking at is the left index finger on top of the head down through the hair, the left shoulder blade, the left buttock, and the left calf. You begin to see changes between one and two. Material was added to the top edge over the hand and a great deal of material was added to the upper back. The rounded left buttock has been sliced off and the upper thigh is trimmed and flattened. At the same time, material was added over the left thigh and calf, so that’s the yellow line representing Back 2 is higher or closer to the scanner than the blue of Back 1. The bottom ledge still appears untouched. If you do the same thing as a horizontal cross section, the features here are the left background, the left standing leg, the bent right leg and the right background. Again, the side edges and backgrounds are nearly identical in surface contour but the differences are that the legs have been widened and certainly on the standing left leg, a large hunk has been chopped off. Now what I don’t know is whether Matisse sort of covered the whole leg over and then chopped off and then re-carved or whether he cut off that slice of the left leg. Beforehand, I tend to think he made it larger and then went and carved it over. And perhaps most striking is this area of the left background, through the figure at the shoulder level, to the right background. The backgrounds of course are similar again, but large differences in the figure show you that the entire scapula and neck area have been covered with a thick layer of plaster to create that smoother more abstract form of Back 2.

Third, laser scanning gave us the ability to measure the relative depths of the sculpture using an inspection program. So, we’re going to find out not only where Matisse made changes but by how much. Here we’re comparing Back 1 to Back 2 and the scale in this view is plus or minus one inch. Green means that areas on one and two are nearly identical in their distance from the scanner up to one inch. Grey areas have a difference of greater than one inch. We notice that Matisse added to the figure heavily but left the background pretty much alone. Variations in the background have to do partially with sculpting and also with waviness in the cast because when you cast a large six foot by four foot thing like this, you do develop waviness in the bronze. Where you see blue, it means that Back 1 is up to one inch lower than Back 2 and where you see red, it means that Back 1 is up to one inch higher.

If you change the scale to plus or minus four inches, different information appears. Again, green means that one and two on this scale have less than a four inch difference. Well no surprise and grey areas indicate a difference of greater than four inches. For example, the standing left leg and the breast on top of the breast on the left are completely new in Back 2 and where you see blue, it means that that Back 1 is four inches lower than Back 2 and where you see red, it means that Back 1 is four inches higher and there again, on the left buttock we know that he sliced that off and this is further proof.

If you superimpose all four backs, you can see where and how the figure changed over two decades. It became wider and it became flatter. Much of the background was preserved all through the series. Laser scanning does confirm that the Matisse Back 1 through 4, are indeed one thing after another.

Speaker Bio

Lynda Zycherman studied at the Conservation Center, and at the Metropolitan Museum with Pieter Meyers. In 1975 she became Conservator at the Freer Gallery of Art, working with Tom Chase.  Her specialty was the technical examination of ancient Chinese material, especially the techniques of ceremonial bronze manufacture. With Elisabeth West FitzHugh, she discovered, identified, and characterized two, hitherto unknown, artificially produced Chinese pigments, Han blue and Han purple. Lynda joined the Sculpture Conservation Laboratory at the Museum of Modern Art in 1984. She has researched a wide variety of topics including Minimalist sculpture, sculpture utilizing electric lights, Fluxus, and Brancusi’s bronze sculptures.

Joe Nicoli graduated from Beloit College in 1996 with a Bachelor of Art Degree in Anthropology. Before joining Direct Dimensions earlier this year, Joe worked all around the Southwestern United States. He worked on archaeological digs for the Navajo Nation and worked on a ruins stabilization crew for the National Park Service. In the last 12 years, he has laser scanned and mapped over 100 archaeological sites for The National Park Service and other clients using a variety of scanning and surveying systems. He is an expert in field data capture and the development of 2D and 3D graphics for documentation, conservation, analysis, and education.

Harry Abramson graduated from James Madison University in 1989 with a Bachelor of Science Degree in Economics. With a career in technical sales and project management along with a love and respect for the arts, Harry joined Direct Dimensions in 2004 to develop technical solutions serving the art industry.  Harry’s work has helped countless sculptures to be realized in every scale, material, and price range imaginable for artists ranging from world renowned to local students. Furthermore, Harry has directed projects that have yielded research and/or archival data for Museums including the Museum of Modern Art NY, National Gallery of Art, The Baltimore Museum of Art, the Smithsonian Institution, and many others.

Glenn Woodburn graduated from Towson University with a Bachelor of Science degree in Industrial Design in 2004. Glenn started working with Direct Dimensions as an intern while at Towson was hired full-time upon graduation. With 8 years experience, Glenn serves as a technical project manager specialized in on-site high-resolution laser scanning projects spanning the art, architecture, historic preservation, film, medical, aerospace, military and product design worlds. Glenn has extensive knowledge in all current and emerging 3D measurement and digital modeling technologies.

National Center for Preservation Technology and Training
645 University Parkway
Natchitoches, LA 71457

Email: ncptt[at]
Phone: (318) 356-7444
Fax: (318) 356-9119