Todor Georgiev, a scientist at Adobe, has etched a lens so that it creates many images. "Software merges the mini-images into a single image that the photographer can focus and refocus at will." (Link to BoingBong report, not original, which may be BoingBoinged/Slashdotted.) So that's what we'll be using manycore machines for.
I had always assumed that the compound eyes of insects and mantis shrimps consisted of a gazillion separate lenses and retinas. If I read this abstract correctly, that seems to be true, at least for our friend the mantis shrimp. But that which Adobe can do in software nature can exceed. Some animal eyes are just pinhole cameras (such as, apparently, that of the white crappie). But what if you had two (or a gazillion) pinholes projecting onto the same retina?
The advantage, you'd think, would be that you'd get the higher resolution of a big retina (the data from a single lens being spread out across more receptors) and the high motion sensitivity of a compound eye (from the lenses positioned more obliquely to the object).
Of course, the signal falling on a single rod/cone is much more complex to interpret. You might need a visual cortex the size of a basketball. And the signal is more subtle, too, since you'd want to be able to detect a fluctuation occurring just in one or a few lenses while the total light falling on a given cone/rod would be contributed by all the lenses. Maybe rhodopsin / existing photoreceptor chemicals wouldn't cut it.