|
|||
The future is now, or at least getting there. They are actually working on doing exactly that, at least with voices. By taking on the order of 40 hours of spoken material from a given person they are working on a technique where fragments could be pieced together to create whatever spoken speech you want with that voice. I'm a little skeptical that they would have the ability to control intonation and expression this way; they might be able to create the word sequences they want (or even to synthesize words which are not among those recorded) but controlling emphasis would be harder. Still, it's not impossible and they're just getting started. There have been several movies now which were completely computer-synthesized, most famously "Toy Story". With the release this year of "Final Fantasy", though, that process has taken a substantial leap because this is the first time that they've actually tried to create realistic looking protagonists. No-one will be fooled, but it's an amazing achievement and if you look how far they've come in just ten years it becomes evident that within another ten it will be possible to create synthesized movies indistinguishable from reality. Actually, in some areas they are doing those kinds of things now; computer-synthesized special effects are nearly ubiquitous in action films nowadays, and it's become the tool-of-choice for many kinds of TV advertisements. In some cases it's become quite difficult to tell where reality ends and synthesis starts. But until now computerized animated films have always used real human voice-actors. Read the credits for "Toy Story" and you'll see Tom Hanks and Tim Allen. Look at the credits for "Shrek" and you see Mike Meyers and Eddie Murphy. Will that be the case in 10 years? One of the problems that producers of family-story sit-coms have is that child actors grow up. Over the course of a five year run, adult actors don't really change all that much, but kids will change quite dramatically and that cute four year old becomes an obnoxious nine year old. But not if they are synthesized. New Shirley Temple movies, anyone? There are two future steps to watch for. First will be the emergence of a TV show where the actors are all digitized from real humans. The humans will come in and be paid for for an extended metric session, where their bodies and movements are digitized and their voices captured. They'll then be handed a big check and no longer needed. In this, as in so much else, new technology will demand new law. I think what will eventually emerge is a licensing scheme where the original actor will be paid a royalty each time their synth is used, and may be able to specify within limits what kind of material the synth can be used in e.g. "no porn". This will lead to the second step: the emergence of a completely synthesized star, not based on any specific human at all. There's already been interesting work where pictures of a large number of people's faces were captured and averaged, to create a synthetic face not belonging to any of them. Interestingly, if you start with forty faces and do this, the resulting average face is considered "good looking" by people no matter who the original forty faces were. After a reasonable number of these actors have been digitized, someone will average them to create a completely new synthetic person unlike anyone who actually exists (except, perhaps, by coincidence) and may do the same thing with their voices. Or perhaps they'll just hire forty people off the street to come in and be digitized. Issue a casting call for forty Chinese women and create a generic beautiful Chinese synth actress. Or forty AA men. Do the same thing three times with the same ethnic group and you'll end up with three synths, all different, all attractive. And such a completely synthetic actor could make movies essentially forever without aging or being hurt by doing his own stunts. Of course, advances in the state of the art could make a given version of the mesh obsolete, but you'd retain the original film used for the digitization and could create new meshes as necessary to keep up, while retaining the distinctive look of your synth. Makeup and costuming are obviously no problem, and if you need to age the person for a part, you no longer mess with rubber protheses; just apply an aging-algorithm to the mesh and you're set. (For that matter, with proper analysis it might be possible to do the opposite and take an adult and produce a child version.) If you've got a female synth and she isn't built right for the part you've cast her into, just twist a dial and grow the relevant parts. No more need for falsies or creative support clothing. Need a hunchback or a scar? Does the part require an amputation? It's all just ones and zeros. Want to cast your star into an ethnic role they don't match? No problem. Take 
| |||