AI Paper Atlas

Research map / Video / Video generation

Geometry-Aware Single-Image 4D Synthesis via Dense Trajectory Generation

Yanran Zhang, Ziyi Wang, Wenzhao Zheng, Zheng Zhu, Jie Zhou, Jiwen Lu

ECCV 2026 · Video › Video generation · approach: Trajectory-guided motion control · 4 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 depth-guided motion normalization (ablation)4D-STraG · VBench Group IIIAesthetic quality: 0.4672 -> 0.4820Improved
add depth latents as conditioning (ablation)4D-STraG · VBench Group IIIDynamic degree: 0.8500 -> 0.9000Improved
add Motion Perception Module (ablation)4D-STraG · VBench Group IIIDynamic degree: 0.8500 -> 0.9000Improved
use patch-level motion features instead of global CLS token (ablation)MPM · VBench Group IIIConsistency: 0.8743 -> 0.8702Mixed
train on 60k filtered samples instead of 1k subset (ablation)4D-STraG · VBench Group IIIConsistency: 0.8567 -> 0.8702Improved
filter training data instead of using unfiltered 60k samples (ablation)4D-STraG · VBench Group IIIAesthetic quality: 0.4654 -> 0.4820Improved
add random noise to training trajectories (ablation)4D-STraG · VBench Group IIIConsistency: 0.8702 -> 0.8685No clear effect
use geometry-conditioned trajectory generation instead of Gen3CMoGe4D · VLM 4D consistency Group IIIAverage VLM score: 2.22 -> 3.76Improved
MoGe4D vs 4Real on subject consistencyMoGe4D · VBench Group ISubject consistency: 0.9329 -> 0.8752Hurt
MoGe4D vs DimensionX on motion smoothnessMoGe4D · VBench Group IIMotion smoothness: 0.9827 -> 0.9760Hurt
use direct trajectory generation instead of generate-then-track pipelinesMoGe4D · Group iiTrajectory error: 0.197 -> 0.042 (Gen3C)Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 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