Recently I posted my new video called Secret Society of Soul Painters. It was created as an experiment in using artificial neural network for generating visually interesting image sequences that would accompany my music. I used Linux based framework Torch and neural-style code by jcjohnson@Github (at this point I would also like to thank hughperkins@Github for his contributions to neural-style project).

Overall it took about a month to prepare a workflow, prepare source and style images, render out the frames and put them back together. I run into some problems with rendering, I described them on neural-style issues Github page. Also I could output only 720p in a final video due to limitations in GPU’s video memory I had available.

There appear to be some interest in more in-depth walkthrough of the process so I will try to deliver a video on that soon. [Edit: Here it is for anyone interested in the workflow]

Enjoy Secret Society of Soul Painters here.

Video was featured in articles on: Fusion.net, Fast Company, Design Taxi, Huffington Post, Prosthetic Knowledge. I very much appreciate this, thank you!