A Form-fitting Photovoltaic Artificial Retina

I still have a problem with this design - I know the chip can make sense of the video input - or even transmit a copy to some sighted person, who can see it as a video - but where is the translation from video format into "human eye-neural signals" format for the treatment of blindness?

There is an explanation given here, that the image is encoded as "light or dark" spots. The signals are sent as "pulses" to an electrode array in the retina - 16 or 60 electrodes in the prototype. I quote from the HowStuffWorks article:

"It takes some training for subjects to actually see a tree. At first, they see mostly light and dark spots. But after a while, they learn to interpret what the brain is showing them, and they eventually perceive that pattern of light and dark as a tree."

It will be interesting to hear some first hand accounts of how useful the stimuli are, whether there are any side effects etc. from the volunteers in the upcoming clinical trials. Apparently the handful of volunteers who have used the prototype for 4-5 years have developed ways of using the spotty input for some benefit.