In the last few years, generative machine learning models have been rapidly improving and changing the landscape for creating images. Abstract concepts like “style” and “content” of an image can be boiled down into statistics using algorithms made possible by advances in the processing power of GPUs (graphics cards). At the same time, Google has made access to GPU resources and Python notebook environments completely free on the web through Google Colaboratory. This combination is enabling anyone, even without any coding or technical background, to experiment with and use these models for their own creative works.
In this workshop, participants will:
- Learn to open and run google Colab notebooks.
- Learn how to upload images to Colab and use them in pre-made notebooks for Style Transfer.
- Learn about CLIP, a model that combines both images and text descriptions to generate or modify images.
- Use CLIP (also in a colab notebook) to take existing images and guide them towards your vision with written descriptions.
- Use CLIP to guide images towards other images, enabling another form of style transfer.
- Experiment with different combinations of images and art to create something awesome!
No previous experience with coding or Google Colaboratory required! Participants will need a computer with an internet connection and a browser.
About the workshop instructor: Miles Vesh Miller is a digital artist and Machine Learning Engineer at Snap. He currently resides in San Francisco.