Patrick Tresset and Frederic Fol Leymarie of Goldsmiths in London are taking NPR to a new dimension… ours. Instead of remaining within the confines of a computer / monitor configuration, they literally bring the computer to the table by means of a mechanic arm. This is not only clever – traditional artist-model roles are upheld – but also meaningful. The portrait situation hits the nerve of perception and vision studies and I hope that future work will explore this more directly.
portraiture as dynamic process
The current portraits take a snapshot of the model, process this visually and transfer the results to commands which are carried out by the robotic arm. Note: this is my guess from the demonstration videos. This sidesteps two of the most interesting aspects of artistic process. Namely, the portrait is a dynamic process. The dynamic quality of sampling information over a period of time instead of a captured instance is an integral difference between photography and artistic perception. The model tries to hold relatively still and the artist tries to piece together numerous glimpses into a figuratively recognizable likeness. The discrepancy is a great source of artistic interpretation. I see this in the videos. The models pose, get photographed and then fall into the role of detached observer. They immediately recognize that they are no longer being artistically scrutinized. The tension dissipates. I would love to see this very interesting project expanded by a dynamic element, having the ‘eye’ continuously update its sampling, piecing elements together. The resulting errors would of course lessen the exactness, but heighten the tension – and thus come much closer to replicating portraiture.
one eye or two?
Another keen element of the vision process is that the artist (or at least most artists) have two eyes. While processing visual input the mind decides between the two, often reflecting human bias. This results in a tendency to ‘unfold’ the face, so that more of the far eye is seen behind the nose than would be possible from a sole vantage point. I often mimic this with mesh deformation or manually controllable rigs that allow a similar deconstruction of a character’s face. It would be fantastic to discover more about the artistic processes behind this bias. Can it be replicated by fragmenting the face into zones and processing scrambled viewpoints to reconstruct the whole? It would also be interesting to see what effect this would have visually.
Head over to the AikonII project site and check out some very interesting research. Its great to see NPR projects that get up from behind the computer and explore the artistic elements beyond algorithms. From their nodal structures and the sensibility shown in their work, I suspect Patrick and Frederick are well on their way.