We’ve had a few posts (this one by Chelsea Kim and this one by Christy Ching) on creating 3D models using photogrammetry, and I thought I’d give some examples of what we are doing with that data once it’s collected. For some objects we are creating ortho-mosaics and these 2D images are going into reports as after treatment images as well as going into the catalogue model as record photography that also shows up in the online collection database. This wooden coffin 2017-20-1.3 is an example of this type of imaging.
For other objects we are also producing ortho-mosaics, but they are before treatment images. For example with E641 a wall painting that was previously on display.
The wall painting is currently in two sections and each one has been imaged separately. These before treatment images have been used to create condition maps.
The maps go into our reports and help provide visual documentation to support our written reports. For large objects, these kinds of condition maps are often easier to understand than written descriptions and can provide more precise information on the location of specific condition issues. Here you can see the condition map for E641. The map is not yet complete, I am still working on documenting one of the sections but I have combined the two maps into one image so you can see what that process looks like.
The models can also be used to show surface distortion, so here in this screen shot of the 3D model of E641 you can see planar distortions in the wall painting where the fragments are not aligned. There may be a variety of causes leading to this distortion including poor alignment during the previous reconstruction or they may be the result of lifting/separation of the original material from its current modern backing.
I am currently working on learning how to create a 2D false color image where the colors reflect depth, so that we can have these planar distortions documented in 2D as well as being able to see them in the model.
So all together, this data is being used to document both the final condition of objects after treatment, as well as to document them before treatment. The models are also useful tools to assess complex condition issues and are valuable for evaluating next steps. For example, our current plan is to remove the wall painting from it’s current modern backing and put it on a new one. Our hope is to correct some of these planar distortions as a part of that process, and this model as well as one we make after treatment will be useful for evaluating the efficacy of the treatment and provide a base line for assessing its condition in the future.
Typically, at the Penn Museum when we are working on objects, even for display, we prioritize stability over aesthetics. This means that we are often do less cosmetic work than would be done at an art museum when it comes to putting in fills and toning out areas of loss. However, I recently undertook a project where I went further than I usually do to recreate lost material. This blog post is going to walk through why that decision was made in this case as well as some of the mysteries that I found along the way
The object in this case is an Egyptian cartonnage mask E1019. When it entered the lab it had a lot of condition issues, including the top of the head was partially crushed, it had been heavily treated before, and it was missing the inlays for its eyes and eyebrows. The missing eye inlays had been giving many visitors to the lab the creeps as the mask appeared to have dark empty eye sockets. Because of this, from the start I had been polling to my colleagues about what level of repair I should do to reduce the distraction of the missing inlays. I was not at this point considering replacing them, but was instead thinking about maybe toning out some of the other losses on the cheek to draw less attention to the eyes.
When it first entered the lab the mask was being tracked as E17632 but over the course of the treatment, I found a different accession number on the interior, E1019. With the help of our curators, we were able to piece together that E1019 was the original accession number, and E17632 had been assigned to it later. When I looked up the record for E1019 in the museum collection database, I found the record included two eye inlays! I was so hopeful that this would mean that I could reintegrate two inlays, one into each eye. However, when I reached out to the curators to get more information, I found out that they are two parts of the same eye, the white part of the eye and a pupil/iris.
Well, this left a new set of problems. Especially since you can see here, the white part of the eye was not very white anymore since it was covered with a dark brown substance. I was left with a lot of options, leave the eye inlays out, reintegrate them as they are, or clean them and reintegrate them, and if I reintegrated them should I then also create a replica set for the other eye?
Before making any decisions, I checked to see if they inlays fit the eye sockets in the mask, which they did. The inlays turned out to be for the masks right eye. After that, I spent some time characterizing the dark coating on the white part of the eye inlay. This included UV examination and comparing how the coating fluoresced with the brown modern materials I found on the interior of the mask from previous treatments. The results were not as clear cut as I was hoping. It seems that there is more than one brown substance on the inlay based on the UV examination. With this data in hand, I reached out again to the curators with the options of leaving the eyes out, reintegrating them as is, or cleaning and reintegrating. The curators indicated that they wanted the inlay reintegrated, and that they would like a replica for the missing inlay as well so that she looked even as one eye seemed worse than no eyes. Together we decided to clean the eye inlay, but to keep samples of the substances on the inlay for future analysis.
Once clean, I set about making a copy for the masks left eye to be a close but not identical match. Based on previous experience I decided to make the new inlay set out of a two-part light weight epoxy called Wood Epox as it is easy to shape and can be sanded and carved. To start, I made a paper template of the shape of each inlay. I made sure to mark what I wanted to be the front of each so that the shape would be a mirror image of the original inlay. The white inlay is slightly curved, so I also created a form that would have the same curvature using foam.
Next, I rolled out some sheets of wood epox, and using the paper template trimmed out the shape I needed for both parts of the eye. The pupil/iris part I let set flat, what let the one fore the white of the eye set in the form I had made so that it would have the same curvature as the original. Once cured I sanded them to finish, with the final stages being wet sanding so that the replica inlays would also have a natural gloss.
The final step before assembly and placement in the mask was the paint them to resemble but not exactly match the originals. I used gloss medium for the pupil/iris as this inlay was especially glossy and I could not get that level of gloss with polishing and painting alone.
Finally, here you can see the end results after treatment. You will see though, that I have not attempted to recreate the inlays for the eyebrows. Because we had the one set of eye inlays, I had something to reference for making the replica set of inlays, however, there are still pieces missing which I had no frame of reference for. There were also likely inlays that went around the outside of the eye as well. These and the brows might have been made out of a variety of materials and without the originals for reference, there is no way to be certain about what their color and appearance would have been.
As an intern working with the conservation department, I have received the opportunity to work on many projects and experience things I never thought I would. Recently I have been working on this software called Reality Capture using photogrammetry. Photogrammetry is a process that uses an abundance of photographs to create a 3D model without any distortion from many overlapping images stitched together to form a detailed and geometrically corrected image called an orthomosaic. This process is usually used on larger objects, and this is because it’s too big to be in frame when taking pictures and has a lower quality with distortion which is far from perfect, and Christy Ching explains more in depth about this in her previous blog post.
I want to show how to create a digital three-dimensional model using the software, Reality Capture, and I’ll demonstrate with an example of the after-treatment photos of an Egyptian coffin.
To start off with, having pictures of the object is a must. For this example, they were already taken and edited in Photoshop, to adjust the white balance using adobe bridge ahead of time. Then I begin by opening the software and then under workflow at the top left corner, I select “inputs.”
Then I select all the images making sure that they were a .jpeg file and then I click on “Align Images” as highlighted above. After the images are aligned, a transparent box surrounding the coffin appears. I adjust the box by dragging the control points around to make it as small as possible without cutting off any part of the coffin. As you can see in the image below, using E883C, the box is close to the coffin but does not intercept the coffin itself.
Now for the fun part to see the coffin take shape, I click next to “Calculate Model” to select “Preview Quality” as highlighted below. Then I go to the tools bar to use the lasso option to erase all the unnecessary space around the coffin. Then after being satisfied with selected area, I click on “Filter Selection,” which turns the selected areas from orange to dark blue showing that it worked.
Finally, I go back to the Workflow bar to select “Texture” which is highlighted below and then it shows all the details of the 3D model without any distortion in high detail and quality.
The conservation department recently acquired a new light for multi-modal imaging – an ADJMEGA PAR Profile Plus (one for use at the conservation lab annex and one for the museum main lab). The MEGA PAR is a tunable LED light source, with 64 different color channels. While not designed for analytical imaging, it provides a bright and large spot size that we can use for visible induced infrared luminescence (VIL) imaging of Egyptian blue. It will also be something we can use to test out other imaging methods in the future. Taking VIL images is not new to the lab, but the light source we had been using stopped working and we needed to replace it. We are grateful to Bryan Harris for making the purchase of the new equipment possible.
Along with the new light, we also acquired a new reference standard, a 99% reflectance spectralon. This standard is critical for developing methods and standard procedures for imaging in the lab. In this post I am going to show an example of how this standard can be used and how I developed a protocol for VIL imaging with the MEGA PAR light.
Since the MEGA PAR light is new, one of the first things I did when it arrived (after unpacking it and reading the instructions of course) was run a variety of tests on known reference materials to see what settings might work for creating visible induced infrared luminescence images of Egyptian blue. As part of that process, I set up a grey scale card (QP card V4) and two reference pigment samples, Egyptian blue and Egyptian green (both from Kremer pigments). I chose these so I would have a known pigment that should luminesce, the Egyptian blue, and one that should not, the Egyptian green. Using the department modified full spectrum camera, I took a visible reference image of the known pigments and the QP card using our regular fluorescent photo lights and a visible bandpass filter over the camera lens so that I could have a normal color image.
Then I captured a series of images using the same set up but replacing the visible band pass filter with an 830nm long pass infrared filter so that I could capture images in the infra-red, with the fluorescent light turned off and the MEG PAR turned on. Each of the images I captured were with the same settings on the camera and with the MEGA PAR light in the same position, just going through each of the 64 color channel options.
I converted the images to grey scale adobe camera RAW by sliding the saturation level from 0 to -100, so that the red, green, and blue values (RGB) would each be the same. I then used the dropper tool to take a reading over where the Egyptian blue standard is in each image and recorded the number. The higher the number, the brighter the luminescence.
After doing that I had a reduced set of options that produced good luminescence in the Egyptian blue for a second round of testing. For round two I did the same thing with the more promising group, but also included in my images the 99% reflectance spectralon standard so that I could check and verify that the light is not producing infra-red radiation. If there is any infra-red, than the 99% reflectance standard should be visible. None of the second round of options showed any infra-red. While any of them can be used for VIL, CL08 gave the strongest response.
After developing a working set-up, I did a test in the photo studio using an object that I knew had Egyptian blue, and the standards. I captured a visible image with the modified camera with the visible band pass filter and the fluorescent photo lights, and a VIL image with the 830nm long pass filter and the CL08 setting on the MEGA PAR. The false color image was created by splitting the color channels on the visible image in photoshop, discarding the blue data, and putting the VIL data in the red channel, the red visible data in the green channel, and the green visible data in the blue channel. As you can see the spectralon is not visible in the VIL image meaning there is no IR radiation being produced by the MEGA PAR light.
After all this work, I had an opportunity to see how the new light would perform in less than ideal settings. I have been working on a study of one of the coffins in the collection, 2017-20-1.3, to examine the coatings and pigments. VIL is the perfect method of identifying blue areas on the coffin but the coffin is too big to fit in the department photo studio. The set of images below were taken in the Artifact Lab (our public lab in a gallery space) where there is IR from the windows (daylight) as well as from the gallery lights. I hoped that a short exposure with the new very bright MEGA PAR would reduce the effects of IR in the image. As you can see in these photos below, the 99% reflectance spectralon is slightly visible but not as clearly as the Egyptian blue on the coffin. These results are much better than what we used to get in the Artifact Lab using our old light, so I am very happy with these results.
One project I have really enjoyed working on as a pre-program conservation technician is documenting larger objects for a process called photogrammetry. Photogrammetry is a technology that gathers spatial and color information of an object from multiple photographs to form a geometrically corrected, highly detailed, stitched image called an orthomosaic. Essentially, photogrammetry creates a distortion-free, three-dimensional model of an object based on two-dimensional photos of every surface photographed in sections.
This can be done for objects of any size. However, we are mostly reserving this technique for larger objects, specifically larger textiles and Egyptian coffins. This is because photographing the coffins and textiles normally with a single shot requires a greater distance between the object and the camera in order to fit the entirety of the object into the frame, and doing so reduces the image quality. Not only that, but the camera distortion that is inherent in all photographs will become more obvious. The resulting image will not be an accurate representation of the coffin or textile, which is not ideal for documentation purposes.
With photogrammetry, we can take parts of the 3-D model and use them as high resolution, distortion-free, 2-D images of the object instead.
So far, a little less than ten coffins, a few textiles, a pithos fragment, and a giant granite relief have been documented using photogrammetry. The models and orthomosaic images are all generated by Jason Herrmann from CAAM, and we are very grateful that he is doing this for us! To learn a little bit more about the photogrammetry process, view this Digital Daily Dig here.
This project was made possible in part by the Institute of Museum and Library Services.
The figure you see here, E4893, is an ivory statuette from the site of Hierakonpolis that I am working on as part of an IMLS grant funded project. I have just started the treatment, but thought I would give a brief run through of the initial examination since this is a good example of when and why we use X-radiography in our department to evaluate the condition of objects before treatment.
You may have noticed that the middle of this object is fill, so not part of the object. The fill has some cracks and splits that suggests it is unstable and should be removed. There is no written documentation for when this fill was done or by who, but it’s possible that this was done shortly after it was excavated. The object was accessioned in 1898. Given that the conservation lab at the Penn Museum was not founded until 1966 that leaves a big gap for the possibilities for when this treatment might have been done.
Based on previous experience, I often worry with these old fills that there are unseen things, like metal pins or dowels, lurking below the surface. X-radiography is a great way to check for these types of hidden previous treatment issues. Though in this case, what I found when I X-rayed the object was not your typical pin or dowel.
Here in the X-ray you can see what I found: while this fill did not have any pins or dowels, whoever had done this treatment had decided to reinforce it by putting nails (4 in total) into the fill material. While this makes the figure look like he has eaten a bunch of nails, it is in some ways better news than a pin would be. Pins usually go into the original material, and if they are iron, can rust and expand causing damage to the object. Pin removal can also be risky and lead to damage of the object especially if the pin is deeply imbedded or corroded into place. These nails, on the other hand, appear to be only in the fill and do not look like they go into the original material of the object at all. This suggests that removal of the fill and the nails should be possible without damaging the object. As this treatment progresses, I will follow up with additional posts and updates.
Inspired by our Building Transformation campaign, we are starting a series on this blog called Transformation Tuesday. At least one Tuesday a month, we will write a post featuring conservation projects underway that are contributing to the Penn Museum’s Building Transformation, along with interesting tidbits that we are learning about the collection and the Museum building as we work.
For instance, we are in the process of fully documenting the architectural elements of the palace of Merenptah currently on display in the Lower Egyptian Gallery, in preparation for the future reinstallation of this material on the 3rd floor of the museum. This includes documentation and research on a variety of levels, and just last week,the palace elements were laser scanned** in order to document them as accurately as possible before they are moved.
Black and white targets placed around the Lower Egyptian Gallery as part of the laser scanning process (left) and the scanning of a column in progress (right)
To supplement the laser scanning, we are creating condition maps of each of the architectural elements. These are big pieces, so it requires us getting up on a lift so that we can examine all surfaces of the columns, gateways, and doorways.
Last week, we began by examining and documenting one of the palace doorways. We always thought that this doorway had been up since 1926 like the rest of the palace, but in the photographs from the opening of the Lower Egyptian Gallery in 1926, this doorway isn’t there!
Lower Egyptian Gallery in 1926 (left) and in 2017 (right). Note that the doorway on the far right in the 2017 image (indicated by the red arrow) is not in the 1926 photo.
We know that the Lower Egyptian Gallery opened on May 19, 1926. So when was that doorway installed? Well, when we got up on the lift to examine it from above, we could see that there were some very minor differences in the way that it was installed, but we also saw a scrap of newspaper stuck between the plaster restoration on one side and the wall.
A piece of newspaper wedged between the wall board and plaster restoration and the wall behind the doorway.
And this piece of newspaper has a date on it! It is a scrap from the Philadelphia Bulletin dated to Thursday July 22, 1926. So it seems that they intended on installing the doorway for the opening, but may have run out of time, and installed it just a couple months later. This isn’t a huge revelation, but it is an example of some of the fun investigating that we do as part of our work to reconstruct the history of our galleries and the objects that are installed in them.
Stay tuned for more Transformation Tuesday blogposts!
My first project as a curriculum intern with the Penn Museum Conservation Department involves the conservation of a mosaic column from the Ninhursanga temple site of the ancient Mesopotamian city of Tell al-Ubaid in Iraq (column in digital collections: (B15887.1 – 15887.4). Dated as 2400 – 2250 century BCE, the column was excavated sometime between 1919 – 1924 as a pattern of alternating triangular and diamond shaped shell, pink limestone and shale tiles. Fortunately, the years of resting in the compacted dirt of the burial environment preserved the shell and stone tesserae and maintained their original conformation. The original column interior, more than likely palm logs, did not survive the centuries of burial.
Figure 1 (left): Before treatment image of the four column sections stacked together to make a mosaic column. Figure 2 (right): Reconstructed façade of the Ninhursanga temple of Tell al Ubaid. The columns were originally believed to be outside the entrance of the temple doorway, although it is not certain if both columns were on the exterior or interior of the building. (Both images courtesy of The Metropolitan Museum of Art. (2003). Art of the First Cities: The Third Millennium B.C. from the Mediterranean to the Indus. The Metropolitan Museum of Art: New York)
Without a support, the delicate tesserae were lacking a method of storage and display. Working with the materials available on site and with technology available in the early 20th century, the archaeological team constructed four drums to mount the tesserae into four stackable sections. Although reports attribute the original 3rd century BCE binding medium as bitumen (i.e., asphaltum, tar, pitch) to hold the tesserae in place on the original wooden supports, the 20th century excavation team reconstructed the shell and stone mosaic pieces with a grey plaster. In addition, only half of each drum holds original tesserae, and the remainder of each section is filled with a painted plaster reconstruction.
Figure 3: c. 1920s, on-site with the recently assembled tesserae onto the wire mesh and wood drums. (Image courtesy of http://www.mesopotamia.co.uk/tombs/story/page07b2.html).
X-radiography of one of the drum sections revealed that each drum is constructed as a hollow metal mesh cylinder with wooden caps on each end, and several nails to keep the cylinder together. After 100 years in this conformation, the drums have become problematic for the long-term display of the tesserae. Fluxuations in temperature and humidity, as well as the weight of the tiles and the thick, rigid plaster have caused opposing shifts in the internal structure, leading to the formation of cracks in the plaster and several tiles to dislocate and fall from the support.
Figure 4: X-radiographs of column section B15887.3 detailing the inner drum structure consisting of an open wire mesh and hollow interior. (Left (a)): X-radiograph of the column section side reveals the radiopaque grid pattern indicative of a metal mesh. (Right (b)): X-radiograph of the column section top, revealing numerous nails in various locations that hold the cylindrical drum together. (X-radiographs courtesy of Julia Commander (2016)).
The column, with all four sections, is one of the many objects selected for exhibition in the upcoming Middle Eastern Gallery (scheduled to open in Spring 2018). Therefore, it was decided it was time for the over 4,000 year old tesserae to be removed from the hollow wire mesh supports and then remounted onto a structured, solid support made with materials that will prevent structural damage and be sustainable for its preservation and long-term display in the gallery. Following a treatment protocol implemented successfully on one of the four column sections by Julia Commander (WUDPAC, Class of 2017), I will deconstruct the tesserae from the current supports, clean and repair each piece, and remount the tesserae in their same arrangement to new cylindrical supports made from solid, very dense Ethafoam measured to the exact shape for each section awaiting treatment. Stop by the Artifact Lab to see the progress of the column treatment, which is already underway, or stay tuned for a follow up blog post!
Figure 5: Start of the disassembly of the shell, pink limestone and shale mosaic tesserae from column section 2. Note the color difference of the large painted plaster fill on the left versus the original tesserae on the right. Several breaks are also observed in the inlays (most notably the beige shell pieces) and a layer of dark, brown grime has accumulated on the surface overall.
Moose hair and birch bark. Those are 2 materials that we have not written about on this blog before. But now that we are working in the Artifact Lab on objects for all of our upcoming exhibitions and loans, we are seeing a wider variety of artifacts and materials in the lab.
Views of both sides of a small birch bark and moose hair case (45-15-1328)
The 2-part case (the lid is a separate piece) was purchased by the museum in 1945. It is attributed to being Huron and from Canada.
The case itself is made of birch bark and it is embroidered with moose hair. The intricate details are difficult to appreciate without being able to see them up close. So, let’s take a closer look at the decoration:
Details of the moose hair embroidery, 7.5X magnification
Details of the moose hair embroidery, 20X
The case and lid are edged with bundles of moose hair attached with thread:
Details of the moose hair embroidery, 7.5X magnification
As you can see in the image above, some of the threads attaching the moose hair bundles along the edges are missing, causing hairs to become lost. These areas, as well as splits in the birch bark, will have to be stabilized before this case can go on display.
Check back for post-treatment photos, and visit the museum to see this case on exhibit by the end of May.
There will be a heavy rotation of objects from Iraq and Iran in the Artifact Lab as we work on objects that will be installed in our new Middle Eastern Galleries, scheduled to open in April 2018. Two of the newest pieces to come into the lab (but 2 of the oldest things in here) are these friezes from Tell Al-‘Ubaid, a site located west of Ur in Iraq, which date to the Ubaid period (ca. 6500-3800 BCE).
B15880, frieze of 6 bulls.
B15883, frieze of 3 ducks
These frieze fragments were excavated by Charles Leonard Woolley in 1924 as part of the British Museum/University Museum Expedition to Al-‘Ubaid. They both have been heavily reconstructed, displayed a lot, and loaned several times, so this is not their first time in the conservation lab. Due to some condition issues and because we are preparing them to go on long-term exhibition here at the museum, we have decided to deconstruct the old repairs and reassemble the friezes using materials that we expect will last longer and provide greater protection for the original pieces.
Conservation treatment of the frieze with the bulls began a week ago:
The bull frieze after 2 days of treatment.
One week later, even more progress has been made:
The bull frieze after 1 week of treatment.
Detail of the first bull freed from the frieze, 7.5X magnification. The bulls are made of shell and are in excellent condition.
Prior to treatment, the friezes were x-rayed to provide a better understanding of their construction and previous repairs, and to guide conservation treatment.
A digital x-ray radiograph of a portion of the bull frieze showing ancient methods of attachment (some are circled in red), modern nails (circled in blue), and a large fill made as part of a previous conservation treatment (circled in green).
Check back for updates on this exciting and complex treatment.