Our Miro Board:
Inspiration:
Kurt Schwitters, Han Arp, David Lobser
1st Sketch:
2nd Sketch:
Based on code from Danny Rozin, David Lobser, and Malo Widerspach
Kurt Schwitters, Han Arp, David Lobser
Based on code from Danny Rozin, David Lobser, and Malo Widerspach
The style-based GAN architecture (StyleGAN) yields state-of-the-art results in data-driven unconditional generative image modeling. We expose and analyze several of its characteristic artifacts, and propose changes in both model architecture and training methods to address them. In particular, we redesign generator normalization, revisit progressive growing, and regularize the generator to encourage good conditioning in the mapping from latent vectors to images. In addition to improving image quality, this path length regularizer yields the additional benefit that the generator becomes significantly easier to invert. This makes it possible to reliably detect if an image is generated by a particular network. We furthermore visualize how well the generator utilizes its output resolution, and identify a capacity problem, motivating us to train larger models for additional quality improvements. Overall, our improved model redefines the state of the art in unconditional image modeling, both in terms of existing distribution quality metrics as well as perceived image quality.
Working alone or in collaboration you will create a virtual being. This may be a completely new character, your 3D scan, an existing person (ethically), or something in between. It can be based on the 3D work we've done in class or an entirely different form (2D, puppet, knitted stuffed animal). With this virtual being you will create a short film, an interactive experience, a performance, or a digital intervention that responds to the conceptual topics explored in class. You may give a live demo in class (time permitting) or a presentation on the project with documentation. Your presentation on the final day of class should be 7-10 minutes.
Things to consider:
What form does your virtual being take?
What platform makes the most sense for your virtual being?
How does this virtual being and the project around it respond to discussions and topics we've explored in this class?
What is the best way to present this project to the class in consideration of the time you have?
L System Study. Design and implement a generative/interactive piece based on L Systems. Record a 20-second composition/improvisation using it.
Etienne Jacob - Looping Curves with Lerp & Delay
Dan Schiffman - L Systems, Nature of Code
Luisa Pereira - Relative Duration & Pitch in Grammars
While the sliders at the bottom only control the visuals, the visuals are responding to an L-System that cycles between F and G.
You can find our code here.