Natal's computational requirements
In order to have some idea of the actual cost involved in the Natal accessory, it is necessary to estimate the computational requirements of 3D full body motion capture. So far I've only heard guesses that they are substantial (requires a modern CPU), and I read somewhere (some academic paper) that facial recognition at around 24 fps with the PSEye was not trivial. Does anyone at Beyond3D have a better sense of what is involved here?
My guess is that the sensor data (audio & video) should be able to fit within the bandwidth limitations of USB 3.0 (~400 MiB/sec), if at some point Natal's sensors are separated from the processing package, or if Sony wants to release a PSEye2 with multiple higher resolution sensors.
In order to have some idea of the actual cost involved in the Natal accessory, it is necessary to estimate the computational requirements of 3D full body motion capture. So far I've only heard guesses that they are substantial (requires a modern CPU), and I read somewhere (some academic paper) that facial recognition at around 24 fps with the PSEye was not trivial. Does anyone at Beyond3D have a better sense of what is involved here?
My guess is that the sensor data (audio & video) should be able to fit within the bandwidth limitations of USB 3.0 (~400 MiB/sec), if at some point Natal's sensors are separated from the processing package, or if Sony wants to release a PSEye2 with multiple higher resolution sensors.