The desire to interface with a computer through something other than text input is deep-rooted. In a demo seemingly inspired by Holly from Red Dwarf, a team of engineers out of Cambridge University last week pulled off something of a media coup. Interviewed by the BBC, their Avatar “Zoe” was proclaimed to be the most “expressive, controllable avatar ever created”.
Sorry guys. You might have spent a week recording a Hollyoaks actress* to synthesize facial movements, but Zoe just got pwned by Ira and made to look like something Tomorrow’s World would have showcased circa 1981.
Presented at the GPU Tech Conference by NVIDIA a few days after the BBC interview, Ira’s raw processing power has delivered [IMHO] a far more expressive experience than Zoe from Cambridge. Just jump to 8min 40secs in the embedded video below.
Now, [here’s the caveat] I realise that the processing power used to control Ira is out of the domain of the humble tablet or home PC. By contrast, Zoe uses a very light software overlay to control the facial movements. However, Moore’s Law won’t keep it that way for long; interfacing with a computer system through something akin to Ira is on the a very real roadmap.
The BBC likes gushing over UK tech and sometimes it’s a little too provincial in its perspective. If there really is a commercial/ technological difference that makes Zoe more practical and deserved of the title “expressive, controllable avatar ever created”, then it didn’t come through in the BBC’s coverage.
…and here’s Ira from NVIDIA. Jump to 8mins 40secs (although the rest of the video includes a fascinating look at the history of rendering the human face).
*No, really they spent a week filming a Hollyoaks actress