Design Adjectives: A Framework for Interactive Model-Guided Exploration of Parameterized Design ...
summary
A domain-agnostic framework maps bounded parameter spaces to a learned intent model that runs interactively and learns from a few labeled examples using Gaussian Process Regression (mean only). A guided rejection sampler generates galleries with four acceptance modes — towards, away, nearby, axis — and a bootstrapper seeds exploration. Implementations for materials, fonts, and particle systems run at interactive rates; professionals spend ~8 hours in case studies and novices complete 10-minute tasks, with studies showing users prefer the adjective-guided interface for exploration.
AI-written summary · about
chapters (10)
- 0:00Introduction to parameterized design
- 1:53Design Adjectives framework principles
- 2:37Gaussian Process Regression implementation
- 3:25Interactive design exploration demo
- 4:33Bootstrapper and low-level controls
- 6:04Guided rejection sampler explained
- 7:32Design gallery and user feedback
- 7:54Bootstrapper initialization methods
- 8:42User study results and evaluation
- 10:13Summary and conclusion
related talks
22mEngineering Intuitive Experiences
46mDesigning for Highly Technical Domains
22mThe Future of Design is Still Personal: Reaching Toward True Adaptation
11mGenTune: Toward Traceable Prompts to Improve Controllability of Image Refinement in Environment D...
28mThe Future of Design is Still Personal: Reaching Toward True Adaptation
25mComputer-aided design of advertisement product compositions
01h 09mGenerative Design: Tools of the Designer of the Future
13mInvisible Users, Invisible Interfaces: Accelerating Design Iteration with AI Simulation