AI Paper Atlas

Research map / Video / Video generation

PhyMAGIC: Physical Motion-Aware Generative Inference with Confidence-guided VLM

Siwei Meng, Yawei Luo, Ping Liu

ECCV 2026 · Video › Video generation · approach: World-model video generation · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add iterative VLM-guided motion probing with confidence gating over one-shot VLM (ablation)PhyMAGIC · five-scene subsetmaterial accuracy (%): 60.0 -> 90.0Improved
add confidence gating to iterative probing (ablation)PhyMAGIC · five-scene subsetmaterial accuracy (%) and probes/scene: 83.0 -> 90.0 accuracy; 1.00 -> 0.80 probes/sceneImproved
add best motion template over fixed boundary for one-shot VLM (ablation)One-shot VLM · five-scene subsetmaterial accuracy (%): 60.0 -> 76.0Improved
increase refinement iterations from 1 to 3 (ablation)PhyMAGIC · Driving caraccuracy (%): 29.17 -> 93.75Improved
increase refinement iterations from 1 to 3 (ablation)PhyMAGIC · Swing ficusaccuracy (%): 62.92 -> 90.63Improved
increase refinement iterations from 1 to 3 (ablation)PhyMAGIC · Rolling basketballaccuracy (%): 50.27 -> 90.00Improved
add confidence-driven prompt refinement to CogVideoX (ablation)CogVideoX · human evaluation plausibilityphysical plausibility: 2.58 -> 2.97Improved
add confidence-driven prompt refinement to CogVideoX (ablation)CogVideoX · CLIP similarity (8 scenarios, average)CLIP similarity: 0.239 -> 0.240No clear effect
full PhyMAGIC pipeline (physics simulation) vs CogVideoX* on aesthetic scorePhyMAGIC · Aesthetic score (8 scenarios, average)aesthetic score: 31.86 -> 30.69No clear effect
replace Physics3D with PhyMAGIC for physics-aware generationPhyMAGIC · CLIP similarity (6 scenes)CLIP similarity: 0.217 -> 0.252Improved
PhyMAGIC vs Physics3D on image-motion distributionPhyMAGIC · Image-Motion-FIDImage-Motion-FID: 98.91 -> 94.69Improved
use OpenSora-2 instead of CogVideoX-5B as I2V generator (ablation)PhyMAGIC · inference accuracy gainaccuracy gain (%): +12.2 -> +10.9No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy