Kelsey Allen's 3DSPA Brings Human-Like Physical Realism Scoring to Generative Video Models
VectorInst · x · 2026-09-26
Vector Faculty Member Kelsey Allen explored how cognitive science and AI shape each other at the 2026 Vector AI Summit: current learning systems struggle with novel physical problems, while humans solve them easily via accurate mental simulation.
Her 3DSPA (3D Semantic Point Autoencoder) measures physical realism the way humans do, measurably improving generative video models — a research line worth tracking where human cognitive capacities feed back into learning systems.
More from Multimodal
- Qwen Image 2.1 on an RTX 3060: 45s per Image with Acceleration LoRA Recipe — Artefact_Design · 2026-09-26
- Custom ComfyUI Nodes for Easy Video Editing: Inpainting and Consistent Tracking — solomars3 · 2026-09-26
- Prompting Gemini 3.8 TTS with nonsense transcripts yields lovely "humming to self" audio — fofrAI · 2026-09-26
- POV interdimensional apocalypse video generated with Seedance 2.5, prompt shared — LudovicCreator · 2026-09-26
- MiniMax H3 video gen keeps forcing reference images into exact poses, user complains — Capnreynolds999 · 2026-09-26
- Skyline-Scale AI Video Fools Viewers as Giant Surfaces — Ok-Kick5886 · 2026-09-26