This article describes the experience of looping again the virtuous circle created by mixing painting and Artificial Inteligence. If you don’t know such a methodology yet, I highly recommend to read the previous post
That post ends with the generation of the following images by a neural network. Let remember here that these generations are based on a human-made painting, that at the same time was painted based on other neural network generation, that was also based on a human made painting.
To create new paintings based on the last IA-generated images, some aesthetics decision were made. Because of those images colors are cold and dark, a fluorescent orange underpainting will make the the painting pop and will open the color scale to a wider range.
So, given the first generated image, the following painting is being created. The image correspond to the moment I am writing these lines.
Given the second generated image, the following painting is being created. The image correspond to the moment I am writing these lines.
At this point, I cannot resist not generating, using the neural network, another batch of images, using the previous paintings as input. Those are the results of both generations
I insist here on the idea of the virtuous circle of creativity. We should think a bit about the ammount of new possibilities that we can produce in no time with little effort..
I think it is very interesting to see how most semantic keys vary on each generation, but somehow composition features (directions, weights, space ) seems to be preserved. The only keys that seems to continue generation after generation is “humanoids wearing clothes”.