杏吧原创

DIGITAL faces. Digitised faces. Morphed faces. Composite faces. Face transplants. Working out what a face signifies is an increasingly complex matter in a world where an archive image of Hitler is grafted onto an actor, a digitised Marlene Dietrich makes a comeback in a new film, and Michael Moore grasps George Bush in a wholly improbable embrace.

In a relatively short space of time, we鈥檝e become de-sensitised to the carefully lit, manipulated and touched-up faces we see in newspapers and magazines. Computer-assisted facial animation, virtual reality and online avatars have prepared us for another layer of unreal reality. The combination of advanced biometric techniques, encoding facial characteristics, with photographic and moving image enhancement, has both improved media images and enabled the technology of facial identification to be used in forensic science and security systems.

Above all, is the ubiquity of these changes preventing us from noticing them properly, and considering their implications?

Supposing the destiny of the face lies in the virtual world? What if the face is no longer shaped by biology alone or by cosmetic surgery, but by humans using new technologies to determine its features, contours and shapes? What will people choose? Can fashions like the current ones for smaller jaws, baby faces and full lips significantly affect the way we look, live and evolve?

This is just the start. It may be fortunate that the technology is not yet sufficiently advanced, therefore allowing us space for reflection and debate.

Modelling faces and facial animation will always be complex tasks because we transmit and receive information using our faces, and facial expressions are determined both by the static and dynamic aspects of the face and by emotional and mental conditions.

The first 3D facial animation and parameterised 3D facial model was created by Frederick Parke from the University of Utah in the early 1970s. These were followed a decade later by the first model of the face which simulated the movements of the real human face, created by Stephen Platt from the University of Pennsylvannia. In 1985, a short animated film about a piano player, Tony de Peltrie, made cinema history by using computer facial expression and speech animation as an integral part of the story.

These days integrating animated 3D models into 2D live-action footage is commonplace, as in recent blockbusters Kill Bill and The Lord of the Rings. And the success of ET, Toy Story, Shrek, The Wrong Trousers, Star Wars and The Lord of the Rings shows that convincing and believable 鈥渃artoon鈥� facial animation is already possible.

But creating digital humans remains problematic. While they have been used in crowds, distance shots and as virtual stunt doubles, digital humans still don鈥檛 stand up to close examination.

Part of the trouble is something the old masters knew only too well: skin tones and textures are difficult to execute. As Rudy Poot, former Colour and Lighting supervisor for Warner Bros鈥� The Matrix, remarks: 鈥淣o one鈥檚 ever really been able to do human skin because it鈥檚 so complex. There are so many layers of light being absorbed by our skin and bounced around and it鈥檚 very hard to mimic that in a program.鈥� Hair is also very difficult partly because it has to be modelled strand by strand.

But the hardest thing of all for generations of artists is the movement of facial muscles. Smiles, which have fascinated artists from Leonardo da Vinci to Francis Bacon, still elude digital artists. Bacon once told a friend: 鈥淚鈥檝e always hoped to be able to paint the mouth as Monet painted the sunset.鈥�

But the difficulty lies in producing faithful human-like computerised faces. This is because software based on an anatomical model that shows the complex interplay between bones, cartilage, muscles, nerves, blood vessels, subcutaneous fat, connective tissue, skin and hair has yet to be developed.

The naturalness of the face is also dependent on timing. A smile can be natural only if spontaneity is maintained and the animation of all parts of the face is perfectly synchronised.

Another challenge is producing digital faces that seem both lifelike and can interact with human beings. Eye gaze and movement are particularly complex, because they vary according to whether someone is listening or talking. Even the speed of a blink or a smile can convey completely different meanings or feelings.

Researchers at the Beckman Institute for Advanced Science and Technology at the University of Illinois are trying to capture the subtleties of expression that are so crucial to mimicking human communication. They are building the world鈥檚 first large database of racially diverse 3D scans of human facial expressions. Beckman fellow Jesse Spencer-Smith is writing software that should enable a computer to recognise a user as soon as they sit in front of the screen.

Across the sectors, researchers are borrowing from each others鈥� work 鈥� in films, surgery, archaeology and computer interfaces. For example, digital facial reconstruction used by archaeologists and forensic scientists to re-animate statistical estimations of dead people鈥檚 faces based on their skeletons is a rapidly developing field. One such 3D graphics program is being developed by Kolja Kyhler and Jyrg Haber and their colleagues at the Max Planck Institute for Computer Science in Saarbr眉cken, Germany. These programs are not only speeding up the meticulous process of building up a face from a skull, but are giving the reconstructed face some lifelike facial expressions.

Similar technologies can also digitally resurrect the dead. The hauntingly beautiful 鈥淒igital Marlene鈥� was created by digital animator Daniel Robichaud for Virtual Celebrity Productions, now the licensing and merchandising company Global Icons, which, extraordinarily, aims to 鈥渉elp celebrities protect and extend themselves as brands鈥�. An increasing number of digital cloning systems allow the faces of dead actors to be 鈥済rafted鈥� onto live ones, thereby ensuring that the dead star has a posthumous acting career.

Another species of 鈥渃yberghost鈥� is beginning to materialise. Last year, digital artists from around the world were invited to send in the computer design of their 鈥減erfect鈥� woman, complete with date of birth and measurements, for Miss Digital World, an internet beauty contest. Franz Cerami, who masterminded the event, claims that 鈥渆very age has its ideal of beauty, and every age produces its visual incarnation of that ideal 鈥� from the Venus de Milo to Marilyn Monroe. Miss Digital World is the search for a contemporary ideal of beauty seen through virtual reality.鈥� While pushing technological boundaries, this raises disturbing questions about the potentially regressive representation of women.

Art publisher Taschen鈥檚 recent book, Digital Beauties, which showcases the latest in 3D animation, is also promising 鈥渟ome images are so stunningly life-like it鈥檚 hard to believe they鈥檙e 100 per cent computer generated鈥�.

Then there is a new generation of supersophisticated digital models. Kaya, one of this elite group, was created by Alceu Baptist茫o, a Brazilian animator and special effects artist, who plans to launch her as a virtual star.

They are undoubtedly striking as images, and by contrast with their analogue equivalents, able to exist and replicate without a link back to the living. But what is apparent in both Kaya and Marlene is a certain lifelessness when they move, which is particularly obvious in the eyes and in the plasticity of the face. What is still lacking in the collision between the organic and the technological is the 鈥渆xperience鈥� of a digital face, a bodily or sensory response to the image: the human needs to 鈥渇eel鈥� for the non-sentient. Without this, comparing a human face with a digital version merely highlights the humanity, uniqueness and enigma of the original.

So despite technological developments, the divide between the artificial and the human remains. Baptist茫o has attempted to bridge this divide in one respect. While most digital artists aim for simulations that are as 鈥減erfect鈥� as possible, he decided to give his model some imperfections. What marks Kaya out from the Lara Crofts are freckles, bushy eyebrows and slightly chipped teeth. But is the result just another layer of artifice? Her uncanny animatedness confuses our perception of her place in our world.

And there is another odd transitional moment: early evidence is emerging that we prefer unreal faces to real ones. A long-term German research project at the universities of Regensburg and Rostock concluded that compound faces were considered more attractive than existing faces. Out of 16 faces selected by a model agency, 14 were unreal composites. The most attractive faces, therefore, may well only exist as unattainable morphs.

Why would this be? We must remember that these faces are more symmetrical than even nature could produce. In preferring these faces, are we unconsciously engaged in parodying nature? Perhaps the clue has something to do with the fact that faces have evolved in tandem with our evolution into sentient beings. We still have so little understanding of consciousness or how it animates faces that we may make mistakes about what gives a human face its humanness. We are understandably seduced by the quantifiable, the numbers of biology.

How ironic would it be if the siren call to perfection were to flatten the very diversity we need to be human. Surely this casts Oscar Wilde鈥檚 famous adage in a different light: 鈥淟ife imitates art far more than art imitates life.鈥� So perhaps the more digital Lara Crofts there are in the world, the more our ability to respond to the rawness of real faces is diminished.