Visualization of a 2000-year-old Egyptian Mummy – Peter Morse & Paul Bourke
We first started working upon the idea of visualizing one of the Egyptian Mummies from the Museum of Old and New Art (MONA, Tasmania) collection in 2007, when I was approached by the exhibition designer for MONA, Adrian Spinks – I was working at the University of Western Australia at the time – so a visit was arranged to the Western Australian Supercomputer Program (now iVEC@UWA) to discuss ideas with my collaborator Paul Bourke. This led onto a series of meetings with David Walsh, the MONA curatorial and design teams, and a number of site visits during the construction of the museum.
MONA had arranged for two mummies to be scanned at the Royal Hobart Hospital, using their new Computed Tomography (CT) scanner, undertaken by the radiologist Andrew Saunders with Gerald McInerney (see: medical imaging). This created two key datasets – the first, a set of DICOM files for the mummy and coffin of Ta-Sheret-Min (Egypt, Late Period, end 26th – 28th Dynasty, c. 664–399 BCE; Human remains, linen wrappings, wood, plaster, pigment.)
This initiated the first visualization project – of Ta-Sheret-Min – which took place during 2008-9. We looked at a wide range of exemplars of mummy visualization at a variety of international museums, coming across some pretty impressive examples of volumetric visualization using a variety of different techniques. However, none of them appeared to be developed to the very high level resolution we sought to achieve in this project. This initial enterprise produced a lot of work and crucial insights into how to develop visualization techniques appropriate for the project, as well as some unusual ideas.
An initial proposal for the exhibit was to create a special type of hologram. At the time a new synthetic holographic technology had been developed which was both full colour and supported animation as a function of the viewers’ position. Paul Bourke created several “panoramagrams” of Ta-Sheret-Min, displaying holographic animation as the viewer moved their point of view left to right across the hologram surface. The view gradually reveals the interior of the mummy, from the exterior, through funerary bindings, to the skeletal structure. However, there are currently distinct constraints in resolution and scale for this material- and compute-intensive process. Similarly, the volume dataset can be realised in laser-etched crystal, or it could be printed using 3D rapid prototyping techniques.
In these instances data visualization asks very interesting questions about portraiture, ethics, remembrance, resemblance and commodification.
After all, these processes can be applied to any volumetric dataset – not just mummies. For instance, you could now have yourself (alive or recently deceased beloved family or pets etc.) CAT-scanned and made into a hologram, a crystal paper-weight, or a life-size volumetric 3D prototype replicant. In the future will we be able to print bodies or organs using stem cells or some other medium? Surely, it’s all a question of ‘resolution’ and developments in materials science (amongst many others.) We were not tasked to explore these – any further, anyway. But it’s certainly fascinating to speculate within both the realms of the currently possible and the imaginary future. Where could this go? Where will it?
In 2010 it was decided to focus upon the second DICOM dataset, that of Pausiris (Egypt, Ptolemaic to Roman Period, 100 BCE – CE 100; Human remains encased in stucco plaster with glass eyes, incised and painted decoration) – as provenance and identity had been confirmed, and the artefact was a rarer and more interesting one. An added benefit was that the skeletal structure of the mummy was far more intact.
The Pausiris mummy had been scanned in three sections, achieving a higher resolution dataset – yet introducing problems with alignment of the parts, due to registration issues on the CT gurney. Paul Bourke resolved these alignment problems using data processing techniques that enabled the accurate registration of the entire volumetric dataset. These were combined into a complete netCDF file suitable for scientific visualization using Drishti, volume visualization software developed by our colleague Ajay Limaye at the Australian National University Vizlab. It was processed upon specially-built computers for the manipulation of high-rez volumetric data (with 96-128GB of RAM) using Nvidia Quadro 6000 GPUs, with 6GB of texture memory. The final volume after trimming was 512 x 512 x 2400 voxels. This pushed 2010 state-of-the-art GPUs near their current absolute limit – and, indeed, it produced many crashes, freezes and problems. Which were, eventually, mostly resolved.
Of course, with current generation GPUs and volumetric software we would be able to create a visualization that shows significantly more detail and resolution, as well as explore interesting new ways of interacting with the dataset.
After much examination of how best to approach this, Paul and I conducted a series of renders, ultimately selecting what we felt to be the most effective set of parameters, meeting Adrian’s vision of a gradual revelation of the remains of Pausiris within the sarcophagus.
This is something that David Walsh indicated in a number of conversations: an interest in a kind of phenomenology of the body after death. Long after death: 2000 years; a ‘presencing’ of Pausiris, yet as someone who is ‘truly’ dead – unlike that of the Serrano ‘portrait’ of the recent corpse set in counterpoint across the room, the dead eyes staring towards this immortal death. This is mentioned by Walsh in his commentary published on the iTouch (the ‘O’) exhibition notes (indeed, I understand – the motivation behind it all): it’s a question about liminality; at what point does a deceased body move from a point of recent ‘liveness’ to being an artefact?
This question asks its corollary: how can ‘it’ be interrogated and resuscitated and ‘enlivened’? For Pausiris, it’s a question about deixis set up as a denkmal: who was this person? what was his aliveness like? A person like you and I, I suspect: who lived and breathed in another time and who believed different things; he understood the world in ways we no longer do.
Many approaches were discarded along the way, though they afford future profitable avenues of exploration (e.g. stereoscopic3D, different slicing approaches, computer re-animation, holographic visualization etc.) Decisions here involved the selection of transfer functions and colour gradients that elicited as much visible structure as possible within the data and the most effective way of presenting it – at the highest resolution. This was challenging because of the nature of the artefact – the CT scan basically reveals density information, to which are assigned greyscale values that can then be arbitrarily colorised. However, bone and plaster, for example, have quite similar densities, so distinguishing them is quite difficult via graphics processing alone, whereas relevant and interesting structures may be far more visible to a trained human eye. The possibilities are endless, yet we must be discerning.
The selected renders were composited in After Effects, using masks, complex layer interactions and re-timing procedures. I won’t go into detail other than to say that I am now intimately familiar with every voxel and pixel of the data. It ran through dozens of iterations, requiring lengthy processing operations over a period of several months, before we finally achieved a satisfactory outcome – all this stuff needs to be invisible to the museum visitor.
The final movie file is rendered at 4000 x 1500 pixel resolution, suitable for display upon the custom software runtime environment using two HD projectors, optical path folding and rear projection inside the specially engineered projection ‘sarcophagus.’
A technical description of the installation and quartz-composer runtime software can be found here.
My impression of the whole installation at the end of this lengthy, complex and challenging process is that it is a truly collaborative work of art: a collision of technical innovation, artistic insight, design vision and a unique patronage that, remarkably, all came together to enable us to see these beautiful artefacts in an insightful and original way.
There is something poetic about the transubstantiation of the body of a man who died 2000 years ago into a refrangible object of computational light, revealed by the spectrum from x-rays to optical wavelengths, via technologies that, to him, would seem indistinguishable from magic.
Note: Images courtesy of MONA, Paul Bourke and the author, where indicated.