Finding Dragons
summary
A practical hunt for novelty using old scans and simple generative tricks: a pipeline pulls the British Library million images set, converts pages into high‑dim vectors, trains classifiers like a support vector machine and random forest in R to cluster portraits, initials and decorative diamonds, yielding ~90% accurate buckets so manual curation finds rare gems (about 40 “diamond” patterns discovered). Also builds name Markovizer, single‑line audio/visual generators, tile/morphing tools — most prototypes work and surface unexpected results.
AI-written summary · about
chapters (17)
- 0:00Intro
- 6:18Mechanical Curator
- 9:53Image Classification
- 16:20Face Detection
- 17:53Recode Project
- 18:47Art
- 22:45Words
- 26:02Business
- 26:51Sounds
- 28:54Live Code
- 32:33Image Matching
- 34:00Sketch Maker
- 37:52Finding Inspiration
- 39:28Drawing Patterns
- 43:51Bank Tiles
- 47:34Shape Tween
- 51:32Color Transfer
