A publication of the Archaeological Institute of America
(Courtesy of Josh Harker)
Although artist Josh Harker stresses he is "first and foremost a sculptor," he is also a notable pioneer in the field of digital forensic art. Using a variety of the latest software and imaging technology, he creates CSI-style facial renderings using the standard Gatliff-Snow American Tissue Depth Marker Method.
How did you first hear about Meresamun and become involved with the project?
It was really just through news outlets, through CNN and the Internet, my usual checking of the news, and I came across the story. It's the type of thing I've always been interested in. They happened to be in Chicago, which is where I'm from. So I gave Emily Teeter a call to talk about what they were doing and if they considered reconstructions and so on, and that's really just how it started. Emily was very gracious with all the information and about sharing the data files, and Dr. Michael Vannier got a hold of me and sent everything over.
Can you walk me through the steps of creating a digital facial reconstruction?
As I said, I got the data from the CT scans from Michael Vannier. At that point, the first set I got was basically too much information. There was a lot of stuff I didn't need. So I got back in touch with him and he was able to break the information down into multiple sets of files of varying degrees of quality. Some of the files had better areas than others--some had better teeth, one had a better nasal cavity, others had fewer fragments of flesh and stuff on the data. So I took the best of each file and merged them into one skull that had all the best information. We ended up with about eight to 12 images and they were really all in varying degrees; for example, the one with the best teeth was missing a lot of skull information. I don't know how many he could have generated, but the dozen or so that he sent were perfect for what I needed to do, which was to get what I needed from each one and get them back into a single, kind of "best of" file.
So what features were you really looking for?
Really just the cleanest surface. The process I use is based on the depth of the tissue from the skull. So I wanted as clean a skull as I could get. That was important. And then the teeth are very important. Some landmarks on the teeth, and which teeth, and where they're located, have bearing on some features on the face and where the muscles lie. And the orbits of the eyes are important. There's some geometry in there that shows me where certain muscles attach. Another very important area is the nasal cavity and a little feature in particular called the nasal spine, which is a little spine that pokes out through the base of the nose, and that tells me, basically, the projection of the nose, how far it comes out, how long the nose is. I compare that to the shape of the top of the nasal cavity and that tells me if there's a bump in the nose and also the angle that the bottom of the nose follows. So that really gives me a lot of information--that one little particular bone gives me a lot of information about what the nose is going to look like, which is obviously a very big feature on a face and recognizing somebody.
Did you encounter any challenges working with the CT data?
There weren't any for me. And again, this is all credit due to the radiologist and the team that did the CT scan. The first file I got was everything. It had the coffin with all the wrappings, and I tried to go through that, but I would have had to do that manually. I don't know if there was an algorithm or if there's some sort of selection in how that CT data is generated, but Dr. Vannier was able to give me just the skull images. So the challenges weren't there for me to have to manually pick through that stuff. I just took the best of the best--and put it back together.
Harker's Meresamun reconstruction front views (Courtesy of Josh Harker)
Were you ever frustrated not having access to the actual skull?
What's nice about this skull being wrapped and being in this coffin--and just being studied through the scan and never being handled or exposed to the elements--is that all these little features were in there, like the nasal spine. In almost any other case where the skull is that old and hasn't been protected in this manner, all of those features are going to be broken and lost and worn away, and you really start to guess about certain things at that point. So it was amazing to have this pristine information to go off. Everything was there that I needed to make this as accurate as possible. Almost in another way, I'd say it was less of a challenge, just because it was like working with a very current skull in very good shape, versus having to guess about things.
I would have thought it to be the opposite.
For the most part, it was really a great bunch of information. This new scanner is a 256-slice scan. I guess the predecessor was called a 64-slice. Now, I don't know exactly what the difference is in those numbers. But the point is, this is an exceptionally higher resolution than the previous CT scans were. It's amazing what this technology can do and how much information was there. It's funny...as far as working with the actual skull, if they wanted, I could have converted the data into a format to run through a 3-D printer and actually print a copy of the actual skull. Then I could have made a physical reproduction on that with clay. That's a really neat process. But I'm doing everything digitally. When you put clay on a skull, all those landmarks are kind of hidden once you commit to them. So you have to be very particular and very precise about what you're doing--which isn't a problem--but with the digital method that I developed, I can always go back. I can always recheck contours and depth markers and landmarks and make absolutely sure that I'm not getting too far away from the architecture of the skull, that I'm staying very accurate. And with that, just like I can take that data and I could have a skull printed, I can also take the finished data of the complete reconstruction and I could have that milled or printed also. So rather than having to build that up on a skull in clay then mold that, I could do it all digitally and print that out and make a physical reproduction, if that's required by the project.
So back to the process. You have the dozen or so sharp images. What's next?
I get this merged skull that I've created of all the best features of the skull, the clearest and highest quality and 3-D images of that. Using the software I use to do the reconstruction, I can spin it around and look at every angle of it. So I was kind of able to say, this one's got a great set of teeth, and in this one the orbits of the eyes are really good, this one's got a good nasal cavity. And then I can select those areas. The model is composed of a bunch of polygons that together create a shell. The polygons are so small, that's what controls the quality and the resolution--the smaller they are, the more detail you have in the file. I'm able to select a group of polygons I want, based on the quality, and extract those, and then save them as another file, and then I keep doing that with each consecutive file, based on which part I want. That's how I was able to merge all the best parts of all the different files together and get the best skull. It's all the same information, it's all the same skull. It's just some areas were better quality than others, so again, it was kind of the best of the best model that I was able to put together to start my reconstruction.
Harker's Meresamun reconstruction profile views (Courtesy of Josh Harker)
What kind of software are you using?
There are a few that I use, depending on what I have to do with it. The main ones are in the rapid prototyping industry, the design and engineering industries, and also the computer graphics/animation movie industry--some of the brand names are ZBrush, 3D Studio Max, Maya, MeshLab, and Magics--all of them may be used at different points, depending on what I have to do. I'm rendering the image from the 3-D model. Render, what that means, is I'm taking a picture of that 3-D model with a bunch of information added to it, so that some areas look like skin, and so on.
As soon as I have that best-of skull model, I am able to take information based on the gender and age and race, and I choose a set of tissue depth markers. There are a few databases out there that we use in forensic art to determine what the tissue depths are for a particular skull. Again, those are based on gender and race and age. And if there's any information regarding her weight--if she was particularly obese or particularly emaciated--there are some adjustments there, also. But I choose the appropriate tissue depth marker to go with, and I create those in a program called SolidWorks, that's actually another program I use. And then all these depth markers, they're like little plugs, at different lengths. And they do exactly what they're called--they tell me the depth of the tissue in different areas of the skull. And each one of those is labeled and has to be put on the skull in a specific landmark. There are approximately 25, maybe 30, tissue depth markers that go all over the skull. Once I have this skull with all the markers on it, which tells me how deep the different muscles and skin and fat tissues are going to be in those areas, I basically start building up muscle groups and tissues in all those areas, bringing them up to the levels of tissue that the depth markers tell me to go to. Once I have all that specific anatomy built up, and everything's correct--and I'm following the skull the whole time, because the skull is the architecture that is going to really control the shape an the placement of all these things and control what she ends up looking like--once I have all that on there, the final skin is put over it and then I work on the cosmetic features, you know, the shape of the lips, some of the details of the nose and the eyes, and even the arch of the eyebrows--where they start and stop--everything is determined from the underlying skull. The skull is great--there's so much there to tell us where features on the face lie, and even their shapes, and where the start and stop.
That's amazing. I didn't realize you could determine something like the shape of the lips just from the skull...
Yeah, well, there's a lot of information there--the width of the lips, how thick they are, and then, of course, if there's an overbite or not--that controls the shape of the lips. When I started learning about forensic art and studying it, I was surprised too, really, how accurate it can be if it's followed correctly and how much information is in a skull. And the more of these that get done and the more information that's out there, it's a pretty good community of folks that share a lot the information about what they've learned. So they seem to be getting better and better, too, because of the shared information--different experiences of different artists--it seems to help progressively. And again, the process I'm using, the way I do it for my own personal reconstructions, digitally I'm able to go back and make sure that I'm being accurate and make sure that I'm still following the landmarks on the skull. Basically, we call them layers, I can see through the skin or I can turn layers on and off and make sure that I'm in the right area--that I'm not guessing and that everything is very accurate.
So after all the elements are in place, are there any other major steps in terms of finishing the reconstruction?
That's the major part, as far as the construction of it. After that, it would be up to whomever I'm doing the project for, whether we actually print a model of the data, which is what I was talking about with the skull. So if the client wanted a finished three-dimensional model, like for a museum exhibit or something, I would take that data, I would print it, or have it milled on a milling machine, and actually get a physical model of it. And then I would work that surface. We could paint it, we could put wigs on it and eyes in it and all sorts of things. Or, in this case, they're using the reconstruction for a digital kiosk, and they want to use it in the website and so on, so the digital model was more important to them than the three-dimensional model. So I take the finished reconstruction and apply material information to it to describe things like...to make the skin look like skin, I paint her in the programs, so that we get skin tone, we get eye color. I create hair, put some makeup on her. In this case, because we know what Egyptian eyeliner, makeup, looked like, it seemed appropriate to add that. Other details could be added if artifacts had been found with her--we could have added jewelry, things of that nature. So I added all of that information and I did what's called a render, it's a very calculated rendering of all that information based on a given light source. So I place lighting around her, and it gives a very realistic image of what all those selected materials would look like--what her skin would look like and her eyes and her hair and everything. And with that, I can spin that around. We can do turntables and different views of her. So that's the kind of final step in the reconstruction, it's really the cosmetic detailing or painting of the surface of the model to look real.
Is there anything you can tell from the skull about the appearance of the hair?
There's no information on the skull that tells you exactly anything about the hair other than the race. Certainly, if it's the skull of a black African, you're going to have a certain hair texture, coarseness, that you wouldn't have in a Caucasian. But you still don't know exactly how they wore it or how long or short it was or anything like that, unless that hair--what they call hair mass--is still with the skull when it's found. You wouldn't know anything about that. For her hair, I don't like to make any assumptions or guesses based on nothing, so I wanted to make sure anything I put in there was as accurate as it could be. I discussed that at length with Emily Teeter about what the appropriate hairstyle for that time would be, the color, where it would be cut, even kind of the coarseness of it--was it very thin wavy hair, was it straight? And so on. All that was based on Emily's direction. So I was really being led there and taking the information and doing what I was told. Again, to maintain the accuracy of the reconstruction.
How long does the process take from beginning to end?
Once I have the information, it's about a week for me. I would say, on average, about 40 to 100 hours. For Meresamun, it was about 50 to 60 hours of work. It really depends on the quality of the information I get up front. But other than that, it's approximately a week or two of work and then the finishing touches.
Going through your process, did you learn anything more about Meresamun's features than Mike and Emily were able to tell just from looking at the scans?
I would say their findings were very accurate, what they saw in the skull. They were following, basically, what was very evident from just a quick glance--how far apart the orbits of the eyes were and how prominent the cheekbones are in the skull--and it's very evident that she has an overbite. So all of that was there and all of that was incorporated into the reconstruction, and certainly, it's not even forced--that's where the architecture lies and what I followed. In addition to that, I get into more of the anatomy details when I'm studying it. In particular, what her nose looked like. They wouldn't have known that from just studying the skull. That was something that came to light through the reconstruction. The look of her eyes is controlled a lot by the size of the orbits and where the brow ridge is, where the eyebrows lie. Knowing that her eyes were wide set is interesting, but actually seeing the eyes, which are such an expressive and recognizable detail on the face, once they were in there in the context of the rest of the face, that brought a lot of her particular look together. That really can't be imagined just by looking at the skull. Some of those things need to be built.
Did you get any feedback from Emily about the finished piece?
I spoke with Emily many times, and I'm pleased to say her reaction was ecstatic. And everybody that she's shown it to, they were all very excited. She relayed many times how exciting it is for them to see the reconstruction, not only of her, but just the quality of it--there's so much detail I can put into these--and that they can actually see pores on her face. It was an exciting thing to bring this person to life in this way and have a peek at what she actually looked like. There's always going to be some potential difference between her actual face--if we're lucky enough to have a picture of her, which we don't--and what the reconstruction looks like. But it's amazing how accurate the reconstructions are with the processes and techniques that I'm using. In law enforcement, when they find somebody from a reconstructed skull and compare the picture of the person with the reconstruction--if the reconstruction is done precisely and by a qualified and talented person--it's eerie how accurate they are. So I have to say that in my opinion, I really think Meresamun would have looked very similar to this reconstruction.
How did you originally get into forensic art?
It's just so fascinating. I'm a sculptor first and foremost. And I have a very strong interest in science in general and also archaeology and anatomy. So you can put all these things together, I guess. They kind of came to a focal point in what I was able to do. I realized I could do these things and do them probably in a way that other people hadn't--and hopefully help the field, promote the field. So that was exciting for me, just to see where forensic art had come and then knowing that I could get involved and learn about it, which was something that captured me more than I sought it out. It was just there to follow.
Have you worked on any other archaeology- or history-related forensic projects?
I did a historic reconstruction as part of my studies, of an unidentified Civil War soldier's face. As it ended up, there was a photo of him to compare it to, for study purposes, to make sure that it was accurate when it was all said and done. But I'm fairly new in the field. I've done a lot of studying and a lot of work focusing on the historical, archaeological aspect of it because of my interest in it, but I've also spent a lot of time focusing on getting the digital tools and software to be able to work with all of the techniques that I've learned. I would say the Meresamun project is certainly the biggest thing I've worked on and the pinnacle of all these things I've been trying to put together. It's really come together fabulously for this project, and in my opinion, really proved how well it works and how quickly it can be implemented with a project like this when the data is there.
Are you working mostly in digital format now or are you still sculpting?
With forensic art, I specifically like to work digitally and that's really how I would approach any project. But I would absolutely do it in a traditional way if the project warranted it, either in a two-dimensional reconstruction, a drawing, or the traditional three-dimensional buildup of clay on a skull or a skull model. I know how to do those. I've done a lot of them. And I'd certainly go back. It's just the digital--really getting this pipeline worked out with the digital process--offers so much more than the other methods, as far as accuracy and even in far as features, being able to go back and do different things with it. And it's archival--it'll last forever. The information is there to go back to, as I was saying before, you can print these out and make physical models if you want to, it's just that much better of a process and set of tools to use that I don't see really going back to the other methods. These new ones offer a lot more, so I expect they'll find more of a demand in the future.
With forensic art, you also have looked at modern people as well, correct?
Yes. Most of that is usually based around law-enforcement situations, where you have an unidentified victim or where they've simply found a skull or body and they don't know who it is. Again, there's a lot of information that can be extrapolated from the skull. So when they find the skull, they can tell if it was a male or female, the gender, and the age, all that information is readily available. And you can do that reconstruction and once that picture is out there, of course the general public is going to recognize a face long before they would recognize who a skull belongs to.
Is there any difference between working with an ancient skull and a modern one?
Not really, other than many times an ancient one is going to have an intrinsic historical significance. So it's going to be possibly so valuable that you wouldn't want to damage it or even handle it or contaminate it or risk anything to it, any harm. For instance, King Tutankhamun's skull--that's not really something you'd want to take the actual skull, which is still mummified in the sarcophagus, and do a reconstruction on it. It's just far too valuable of an artifact to do that. And with Meresamun, they didn't want to open the sarcophagus and damage her remains or anything. So the ability to make these scans without disturbing any of the remains--and not even having to touch them--is remarkable for what that brings to the ability to do these reconstructions. There is no risk of damage. You can email the information, too. It just opens up whole new possibilities for working with the information and these bodies and remains that you would never have wanted to risk before now.
In terms of your process, though, a skull is a skull, right?
Pretty much, a skull is a skull. The only difference would be, I think, if you went back to pre-Homo sapiens, if you're getting back into some of the hominids, Neanderthals, and so on, there aren't records because we don't have any specimens to make the measurements on, there are no records or data sets for what their tissue depths would have been based on their age or gender. So at that point, the reconstructions can still be done, but you're going back to just a more anatomy-based reconstruction. You're not really sure how thick those muscles are, whether they're all in the correct places and landmarks and attachment points, and you're not sure how big or small those individual muscles would have been. So that would be the only difference--if you're really, really going back in time, then you're going to have some variance there. But to me, a skull is a skull, if you're talking ancient Egypt or China or anything like that, the information is all the same today as it was 10,000 years ago.
Forensic art is just one aspect of the work you do. Can you tell me about your work in general?
I do portrait sculpture. I do figure sculpture. Also, using digital tools, I'm doing, probably my biggest project now is sculptures of knots and tangles. They're forms that I've drawn, probably for close to 20 years, and before the advent of these new software and technologies, they could never be made. They were too complex. And now with this new software, I can sculpt these and I can print them and I can even have them printed in a material that can be used to make molds so they can be cast, and I'm getting them cast in bronze. And I do product design and development, CAD work, and that's kind of the field that introduced me to all this technology, having worked with CAD programs and 3-D printers and knowing machines. I spent a lot of time working with these softwares and technologies and realized that I knew how I could apply them to sculpture or to forensic art or to whatever. So I'm kind of making these bridges of what these tools were developed for and applying them to other avenues.
Eti Bonn-Muller is managing editor at ARCHAEOLOGY.