AI Paper Atlas

Research map / Video / Video segmentation and tracking

How Much 3D Do Video Foundation Models Encode?

Zixuan Huang, Xiang Li, Zhaoyang Lv, James M. Rehg

CVPR 2026 · Video › Video segmentation and tracking · approach: Layered video decomposition and dynamics · 9 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
use temporal video features instead of per-frame DINOv2 features (ablation)3D probe · DL3DVpoint map error / AUC@30: Point 2.814 -> 1.576, AUC@30 0.245 -> 0.558 (DINOv2 vs V-JEPA)Improved
use temporal video features instead of per-frame DINOv2 features (ablation)3D probe · CO3Dv2point map error / AUC@30: Point 0.559 -> 0.439, AUC@30 0.508 -> 0.619 (DINOv2 vs V-JEPA)Improved
add 3D-aware finetuning (Aether) to CogVideoXCogVideoX · DL3DVpoint map error / AUC@30: Point 1.748 -> 1.566, Depth 0.608 -> 0.574, AUC@30 0.486 -> 0.527Improved
add 3D-aware finetuning (Aether) to CogVideoXCogVideoX · CO3Dv2point and depth error: Point 0.485 -> 0.501, Depth 0.231 -> 0.249Hurt
scale WAN2.1 from 1.3B to 14B (ablation)WAN2.1 · CO3Dv2 ablation subsetpoint map error: 0.0468 -> 0.0360Improved
scale CogVideoX from 2B to 5B (ablation)CogVideoX · CO3Dv2 ablation subsetpoint map error: 0.0576 -> 0.0590Hurt
replace DINO features with frozen WAN2.1-14B features in VGGTVGGT · CO3Dv2point map error / AUC@30: Point 0.476 -> 0.289, AUC@30 0.565 -> 0.718Improved
replace DINO features with frozen WAN2.1-14B features in VGGTVGGT · DL3DVpoint map error / AUC@30: Point 2.751 -> 1.034, AUC@30 0.363 -> 0.686Improved
halve probe width from 1024 to 512 (ablation)3D probe (WAN2.1-14B) · DL3DVpoint map error: 1.051 -> 1.317No clear effect
use WAN2.1-14B features instead of Fast3R features3D probe · DL3DVpoint map error / AUC@30: Point 1.379 -> 1.051, AUC@30 0.637 -> 0.660Improved
use WAN2.1-14B features instead of Fast3R features3D probe · CO3Dv2point map error / AUC@30: Point 0.262 -> 0.284, AUC@30 0.769 -> 0.736Hurt
extract features from mid-network layers at early-but-not-first timestep (ablation)video diffusion models · CO3Dv2 ablation subsetpoint map errorImproved

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