Five LLMs, one physics prompt, one ground truth — a blind benchmark that scores AI-generated 3D underwater acoustic ray-propagation panels against a real BELLHOP3D run on data, not pixels. Vanilla JS + raw WebGL, zero dependencies, one shared harness.
webgl benchmark canvas vanilla-javascript static-site scientific-visualization computational-physics sonar underwater-acoustics ray-tracing 3d-visualization bellhop zero-dependencies llm-evaluation ocean-acoustics ai-benchmark llm-benchmark acoustic-propagation transmission-loss
-
Updated
Jul 13, 2026 - HTML