top of page


Creative Coding - Visual Art Focus: Machine learning training & trails

style transfers made in google colab

Next frame prediction

To be quite frank working my through the google notebook colab at first was quite nightmarish. I would watch the tutorials and scroll angrily through trying to understand whats what or where things go. It took quite a while to figure out, oh that's where I upload the files and this is where things come out. Some notebooks were pretty straight forward upload here and results are right there, but others like the deep dream. No I just couldn't figure it out, error after error kept coming and I couldn't understand how to fix it, it couldn't read my uploaded image somehow. So in that case frustration got the best of me and decided after the third day to just let it go. I had the same experience with the NFP, but with this one it was all about patience and trail and error. If the video was too long I learned to shorten it. If the computer slept I just kept a video running so the training wouldn't stop. Finally with this one persistence paid off, but was the result worth it. Well my opinion thus far is undetermined with a mix of pain.

Reference used: Anthony Bourdain's parts unknown, Sir Lanka episode.

Reference: Family photo

Style transfers made in google colab. Reference: Hajime


bottom of page