I used the ml5 hand pose feature to track hand movements and create a kaleidoscope effect. the webcam captures the positions of your fingers, and the kaleidoscope responds by drawing symmetrical patterns based on your hand movements
https://editor.p5js.org/yg2348/sketches/EHTmfl2Ed
The model I found interesting was the AI Portrait project, where an AI artist named "AI Gahaku" generates a masterpiece from your photo.
The AI Portraits project uses a type of machine learning model called a Generative Adversarial Network (GAN). This model has two parts: one that creates images and another that checks how real they look. The two parts work together, making the model better and better at generating images that look like real portraits.
GANs are great for this project because they can create completely new images that look like they belong to a certain style, such as classical paintings or modern art.
To train the model, it likely used a large collection of portrait images from different time periods and art styles. This could include classical paintings, renaissance portraits, and other famous artwork of human faces. By learning from these images, the model can understand the features of different styles and apply them to new portraits.
Because the model has seen a variety of artistic portraits, it can take a photo of a person's face and transform it into a unique piece of art that looks like it was painted by a famous artist from history.
The creators chose GANs because they are really good at creating realistic images, even when starting from nothing. GANs can learn to imitate different art styles, which makes them perfect for a project that aims to create portraits in the style of famous paintings.