Due Thursday, Oct 10: Watch Machine Learning for Human Creative Practice by Dr. Rebecca Fiebrink at Eyeo 2018. Reflect on the question: How can machine learning support and expand people’s creative practices? Write a response to this question and outline an initial idea for your project. Do you plan to build on a previous assignment or start something new? This can be a short post—just a paragraph response and a sentence or two about your idea! Post a link on the ml5.js Project Wiki page.
Machine learning can support and expand people’s creative practices by extrapolating data that’s usually difficult to measure with programming and use that data to control something that might be unintuitive for traditional controllers (buttons, midi keyboards) to do. A classical example of this would be body pose where there’s millions of pixels over hundreds of frames to work with and impossible for human programmers to extrapolate any useful information in real time.
There also happens to be a lot more “happy little accidents” regarding machine learning outputs then traditional programming outputs. For programming, you usually can expect the results of the program you designed, including bugs and exceptions. But for machine learning, I find it to be a lot more “subjective” and results are less predictable as I experiment with different training sets and algorithms. Transferring such unpredictable data into arts can create interesting results like with earlier models that generate psychedelic stylizations of an image input.
For the ml5js project I want to create sort of paint brush with body pose, but much more dynamic than just shapes being drawn at the same location of the hands position on canvas. I’ll try looking into the book Generative Design for more fluid examples of dynamic brushes to map with my hands, and then maybe I’ll train the neural model or use Teachable Machine to recognize “swipe left/right” as changing brush type, pinch to erase everything, and open hand pose to draw.
Here are some examples on more organic, dynamic drawings that could potentially be drawn with my real life hands.
