AI Paper Atlas

Research map / Video / Video generation

PhysChoreo: Physics-Controllable Video Generation with Part-Aware Semantic Grounding

Haoze Zhang, Tianyu Huang, Zichen Wan, Xiaowei Jin, Hongzhi Zhang, Hui Li, Wangmeng Zuo

ECCV 2026 · Video › Video generation · approach: Trajectory-guided motion control · 10 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 cross-attention text conditioning to plain transformer (ablation)PhysChoreo predictor · Ablation: model designMaterial accuracy / total error: 0.5507 -> 0.7393; 3.4230 -> 0.9789Improved
add dual-stage (global then part) cross-attention (ablation)PhysChoreo predictor · Ablation: model designMaterial accuracy / total error: 0.7393 -> 0.8077; 0.9789 -> 0.5801Improved
add soft assignment (ablation)PhysChoreo predictor · Ablation: model designMaterial accuracy / total error: 0.8077 -> 0.8372; 0.5801 -> 0.5046Improved
add segmentation prior (ablation)PhysChoreo predictor · Ablation: model designMaterial accuracy / total error: 0.8372 -> 0.8605; 0.5046 -> 0.3318Improved
add soft assignment (ablation)PhysChoreo predictor · Ablation: model designEarly-stopping iterations: 36 -> 26Improved
add segmentation prior (ablation)PhysChoreo predictor · Ablation: model designEarly-stopping iterations: 26 -> 41Hurt
add L_assign loss (ablation)PhysChoreo predictor · Ablation: loss componentsMaterial accuracy / total error: 0.8534 -> 0.8605; 0.3753 -> 0.3318Improved
add L_assign loss (ablation)PhysChoreo predictor · Ablation: loss componentsEarly-stopping iterations: 50 -> 41Improved
add L_smooth loss (ablation)PhysChoreo predictor · Ablation: loss componentsMaterial accuracy / total error: 0.8578 -> 0.8605; 0.3579 -> 0.3318Improved
add L_con loss (ablation)PhysChoreo predictor · Ablation: loss componentsMaterial accuracy: 0.8310 -> 0.8605Improved
add L_con loss (ablation)PhysChoreo predictor · Ablation: loss componentsTotal error: 0.3451 -> 0.3318No clear effect
use part-level predicted physics instead of global properties (ablation)PhysChoreo pipeline · Video generationPart-level dynamicsImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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