Towards Declarative Content Generation for Creativity Support Tools
summary
Argues that answer set programming (ASP) can specify generative spaces for game/art content but has practical gaps: uses clingo to generate tile‑map heightmaps and contrasts a p5.js cellular automaton with an ASP model, builds a tile generator, narrative planner and a structure editor for ASP. ASP produces concrete sets of valid worlds and catches degeneracies (needs an added global min‑max constraint to avoid “all ocean”) but struggles with sampling diversity and inspecting intermediate states, whereas the CA yields smoother, seed‑tunable evolution that is easy to animate.
AI-written summary · about
chapters (20)
- 0:00Start
- 3:00Hello, PADL!
- 4:36Procedural Content Generation (PCG)
- 6:36Parametric Design and Generative Art
- 7:58Genuary
- 12:24Navigating Design Spaces
- 14:38Possibility Spaces
- 15:40Talk roadmap
- 16:34Motivating problem: procedural terrain generation
- 18:02A tale of two programs
- 21:38Answer Set Programming
- 29:48Solver-Aided Content Generation
- 33:10Usability Efforts
- 34:51Terrain generation in ASP
- 44:45Comparing p5.js/CA with ASP versions
- 48:36Expressive Range Analysis
- 50:33Helping users understand the generative space
- 52:07Proving equivalence of generative spaces
- 53:03Inspecting intermediate states of declarative programs
- 53:53What does it mean to program with "rules"?
related talks
36mPanel: Generative art for 21st century
33mProcedural Content in JS
30mEvolving JavaScript: Cultivating Genetic Algorithms for Creative Coding
57mProcedural Generation: Fun = Feasability + Fairness + Interestingness
39mGenerative Machines
25mTurning Data Into Generative Chaos
28mCreating generative art with Javascript
49mWhat’s Up with Generative Art (GenAI)