|View single post by Keith Clark|
|Posted: Jun 23rd, 2014 10:43 AM||
I wrote to Artem and Sergey last week.
Sergey provided the additional information, and is ok with it being shared:
"I'll tell you a little about the program:
ITCScope born accidentally from the program for designers, which I'm doing now. I noticed a person, buildings, animals during testing program and sent Artem Mikheev.
A week later decided to share with the community to understand all together, can there be a future for this method.
Program is based on non-linear processes (neural networks with feedback).
The principle of operation is similar to a video ring. The program initiates the random number generator of the input field, is repelled from the generator and adjusts accidentally neurons. Further neural network draws a picture.
Each pixel is input to the neural network, is computed and the result is fed back to the input. This occurs repeatedly.
This method is almost the same method as the video loop, but neural network replaces the camera and the TV replaces the canvas."
He seems to be somewhat excited about the work he is doing, and I'm sure we'd like to see him continue his efforts.
A little bit of language barrier, a friend translates my emails to him, but the connection has been made. Should it be in the best interests, perhaps there may be some collaboration down the road....