3D Printing the Star Carr Pendant

Left – the unmodified STL file of the Star Carr pendant. Right - A UHD 3D print of the pendant at 1:1 scale made using a 3D Systems Projet 3500 HD Max 3D printer at De Montfort University, Leicester.

Left – The unmodified STL file of the Star Carr pendant. Right – A UHD 3D print of the pendant at 1:1 scale made using a 3D Systems Projet 3500 HD Max 3D printer at De Montfort University, Leicester.

In 2015 an engraved shale pendant which is roughly triangular and measures approximately 31mm by 35mm by 3mm thick was found during archaeological excavations at the Early Mesolithic site of Star Carr, in North Yorkshire UK. On one of its faces it has a series of incised linear marks which the team involved in its discovery claim are the earliest known Mesolithic art in Britain. In March 2016 the independent, on-line, not-for-profit internet journal Internet Archaeology published a scholarly paper on the Star Carr pendant (http://intarch.ac.uk/journal/issue40/8/index.html ). Helpfully, in their article Internet Archaeology included a downloadable STL (Surface Tessellation Language) file of the artefact. This is a very welcome initiative and Internet Archaeology are to be congratulated on taking this step which, if applied more widely, may have the potential to benefit a range of disciplines beyond archaeology.

STL files contain 3 dimensional coordinate data in the form of a geometric mesh of the surface of a 3 dimensional object  and can sometimes have other properties attached like colours. It is an industry standard method of transferring 3D shape data in a wide range of disciplines, notably in engineering, product design and architecture. The mesh data can be created in a number of ways from laser scanning and photogrammetry to 3D solid modelling. One of the uses of STL files is to produce 3D physical prints of the objects. This is something that we do quite a lot of in the Digital Building Heritage Group at De Montfort University and because we feel that Internet Archaeology’s initiative moves the debate about the use of 3D digital data of archaeological artefacts on in a useful and interesting way, we thought it might be helpful to look not at the artefact but at the 3D print arising from it and examine aspects of its production and potential use.

Reproductions of archaeological artefacts can have a number of very different uses. These may range from surface geometry analysis, wear pattern and impact detection to artefacts handling for school children and students, substitute display, travelling exhibitions and experimental archaeology. However it’s important to be clear about the pros and cons of 3D prints in relation to reproductions made by skilled artisans out of authentic materials. Good, hand-made reproductions are excellent for giving an authentic haptic experience of the appearance, texture, weight and “feel” of an artefact type, and of course they are often useable, having the same mechanical properties as the original would have had. Hand-made reproductions are in fact new artefacts, they can have most of the material properties of the original and can give excellent service as convincing simulacra for many purposes. But there are aspects of hand crafted copies which are less effective. One of these is that it’s virtually impossible using hand crafting to give crafted copies precise verisimilitude with the original at a very small-scale, for instance in scratches, incisions, indents, wear marks, ablations, corrosion pits and so on – aspects which can be of importance to archaeological enquiry. 3D data capture can (in theory) record these tiny details on the original with considerable precision, as well as the overall form of the object. The STL file of the Star Carr pendant is an example of this and a closer inspection of it can yield some useful insights into the advantages and limitations of this technique.

Sectional view through the STL file of the Star Carr pendant showing the good surface detail and the internal structural anomalies arising from the method of data capture and digital object production. Ideally these internal anomalies should be removed but in this case they were not critical and the part printed without modification.

Sectional view through the STL file of the Star Carr pendant showing the good surface detail and the internal structural anomalies arising from the method of data capture and digital object production. Ideally these internal anomalies should be removed but in this case they were not critical and the part printed without modification.

On downloading the STL file of the Star Carr pendant from (http://intarch.ac.uk/journal/issue40/8/index.html ) the first step was to evaluate its integrity. 3D prints work best with “clean” STL files, that have no errors, consistent orientation of normals, no overlapping edges between triangles, contiguous “water-tight” shells and no internal anomalies within the 3D model. The Star Carr pendant STL file had a “water-tight” external shell and no extraneous shells but did have some unusual internal structural anomalies which tell a little about the method by which the model had been produced. It would be normal to remove these internal anomalies in the STL file for good 3D printing not least because it would then allow an effective re-hollowing of the part to reduce material usage. In a part this small, and using this material the quantities are trivial, the anomalies are non-critical and in this particular case they do not prevent the part printing, although similar anomalies in other circumstances often do and so the file was not modified before 3D printing.

Starr Carr Banner 3&4 small

Left – The Triangulated Irregular Network (TIN) of the model’s shell. Right – A close up of the TIN around the circular perforation at its upper apex. Surface tessellation edge lengths at their smallest are between about 0.05mm (50 μ) and 0.1mm (100 μ).

Viewing the STL file on-screen also allows you to zoom into the object and inspect the TIN as produced by the capture technique which in the case of the pendant has yielded tessellation edge lengths between 0.05mm (50 μ) and 0.1mm (100 μ) for the smallest features. This is at the higher resolution end of scan data used for 3D printing and are justifiable given the small size of the object and shallow depth of the surface features. For this object at this resolution the result is an STL file size of 30.4MB. Our colleague James  Meadwell here at De Montfort University supervised the printing of the STL file without modification on our 3D Systems Projet 3500 HD Max machine in Ultra High Definition (UHD) with a resolution of 750 x 750 x 890 DPI (xyz). The deposition layers have at best a 29μ thickness and the material was an acrylic polymer with a wax support structure. This is a professional, high-resolution 3D printer and a good quality material.

A 1:1 scale 3D print of the Star Carr pendant produced using a 3D Systems Projet 3500 HD Max at De Montfort University, in Ultra High Definition (UHD) with a resolution of 750 x 750 x 890 DPI (xyz). The deposition layers have a 29μ thickness. This level of layer thickness in 2016 is regarded as high fidelity 3D printing – even so note how the layer contours mask the fine surface detail on such a small (35mm wide) artefact at 1:1 scale.

A 1:1 scale 3D print of the Star Carr pendant produced using a 3D Systems Projet 3500 HD Max at De Montfort University, in Ultra High Definition (UHD) with a resolution of 750 x 750 x 890 DPI (xyz). The deposition layers have a 29μ thickness. This level of layer thickness in 2016 is regarded as high fidelity 3D printing – even so note how the layer contours mask the fine surface detail on such a small (35mm wide) artefact at 1:1 scale.

 

The results are interesting. It’s immediately obvious that at 1:1 scale, even using this high-resolution 3D printer, while the overall form of the artefact is reproduced well the surface detail is largely obscured by the fine layering of the 3D print material. This layering is virtually invisible to the naked eye when looking at the 3D print under normal light and is only visible when photographed with a macro lens. Clearly 3D printing such small objects for such shallow surface detail at 1:1 is less effective than it might be.

Left - the 5x 3D print and 1:1 scale 3D print compared. Right - the 5x 3D print created using a Z-Corp 650 printerin resin bonded gypsum.

Left – the 5x 3D print and 1:1 scale 3D print compared. Right – the 5x 3D print created using a Z-Corp 650 printer in resin bonded gypsum. Note that some of the larger incised marks on the surface of the object are now visible to the naked eye at a distance under angled directional lighting, but that the contour lines of the printing are also still visible.

In order to examine whether this limitation could be overcome we scaled the 3D print up to 5x its actual size and printed one side of it. This was carried out using a different machine (a Z-Corp 650 printer) and different material (resin bonded gypsum) in order to reduce the cost of the print. Cost is largely determined by the solid volume of the object, the pendant at 1:1 scale used 0.87cc’s of material, the scaled up version used 97.18cc’s. Here the layer thickness is between 0.089 (89 μ) and 0.102 mm (102 μ) about three times that of the previous model but the scaling is 5x so there is actually an effective theoretical diminution of layer thickness in relation to the object size, not an increase and so one would expect increased legibility of the surface detail. For the 5x 3D print of the pendant the results indicate that this is in fact the case but not in our view to sufficient extent in this instance to be convincing. The surface detail is a little more legible and even at this size (about 15cm across) can to a certain extent be distinguished by touch with the fingertips. The direction of improvement does however indicate that further increase in size and / or use of UHD printing at this 5x scale (with of course its attendant increase in costs) may yield convincing results. Here at De Montfort University we have the ability to produce physical copies in one piece up to much larger sizes (about 0.5m) in diameter but whether such a capability can find any use for the analysis or display of artefacts is not known at this stage.

This ability to easily change the scale of objects in digital 3D is one of a set of key advantages of the medium that can be applied to many objects to examine aspects of them. Were this approach to be developed further this is perhaps one way that 3D printing should be viewed by the research community, not only to try to reproduce high quality copies of precious archaeological artefacts –  perhaps questionable in some senses but useful in others – but as with so many analytical techniques and instruments as a “lens” to bring into focus particular aspects of the artefact which the technique may, with further development, be naturally suited to effectively examining, in this case surface detail, shape and form. We believe that 3D printing may be useful for:

  • Visual display where handling is not necessary – very convincing 3D printed reproductions can be produced of suitably sized copies which are colour matched and physically excellent shape copies. See https://wp.me/p2eM6I-di
  • Experimental testing, for instance creating moulds and formers for casting or shaping in authentic materials like clay, bronze, precious metals, glass etc.
  • Creating scaled up copies of objects surfaces to illustrate or examine surface wear, or impact damage.
  • Testing physical forms where mass and material of the object are not key factors, for instance in fluid or air flow around or through static objects.

We hope this preliminary examination of the use of this STL file of an archaeological artefact may prompt further discussion on what roles, if any 3D printing may have for meaningful contribution to the study of heritage and material culture artefacts.

 

About Douglas Cawthorne

Reader in Digital Heritage at De Montfort University.
This entry was posted in 3D Printing, Archaeology, Archaeology. Bookmark the permalink.