Picasso's terminal; data science and AI in the visual arts
summary
Shows concrete results not promises: deep dream amplifies doglike hallucinations and spawns fractal puppy slugs, fast tensorflow style transfer runs real time on a laptop GPU, pix2pix turns sketches and maps into photo‑like textures after training on a few hundred paired examples, and CycleGAN maps horses to zebras without pairs. Demonstrations include a live webcam cubist mirror and a hacked face‑swap trained on a few hundred Trump frames that works but looks disturbingly uncanny.
AI-written summary · about
chapters (22)
- 0:00Picasso's Terminal concept
- 1:03Music to art
- 4:11Neural networks in visual art
- 6:59Deep learning in computer vision
- 10:10Convolutional networks explained
- 13:25Uses of convolutional networks
- 14:28Open ML communities rise
- 17:15Keras for deep learning
- 18:21Deep Dream as art tool
- 21:11Deep Dream with Google's notebook
- 24:22Style transfer for images
- 26:51Real-time style transfer demo
- 29:14Generative networks for art
- 34:09Text to images with neural nets
- 35:13Pix2Pix sketch-to-image
- 37:18Invisible Cities project
- 40:51Trump face with Pix2Pix
- 44:02CycleGAN for image translation
- 45:49Live image captioning
- 48:18WaveNet for audio
- 48:59ML resources for artists
- 52:31Q&A
related talks
25mWhat if Picasso was a robot? Teaching art to the machine
57mWhat if Picasso was a robot? Teaching art to the machine
31mPicasso, Geometry, Jupyter
01h 03mIt's All About AI
39mKeynote: Artistic Applications of Artificial Intelligence
25mDecoding Algorithmic Artistry: Visual Generative AI
01h 00mData, visualization, and designing AI
36mArt in the Age of Algorithms