AI Paper Atlas

Research map / 3D vision / 3D generation

Inferring Compositional 4D Scenes without Ever Seeing One

Ahmet Berke Gökmen, Ajad Chhatkuli, Luc Van Gool, Danda Pani Paudel

CVPR 2026 · 3D vision › 3D generation · approach: Monocular 3D object reconstruction · 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 static/dynamic embeddings (ablation)COM4D baseline · DeformingThingsIoU: 0.2018 -> 0.4034Improved
add static/dynamic embeddings (ablation)COM4D baseline · DeformingThingsChamfer Distance: 0.1525 -> 0.1284Improved
add static/dynamic embeddings (ablation)COM4D baseline · 3D-FRONTChamfer Distance: 0.1668 -> 0.1044Improved
add Diffusion Forcing to baseline (ablation)COM4D baseline · DeformingThingsIoU: 0.2018 -> 0.4271Improved
add Diffusion Forcing on top of embeddings (full model) (ablation)COM4D · 3D-FRONTChamfer Distance: 0.1044 -> 0.0909Improved
add Diffusion Forcing on top of embeddings (full model) (ablation)COM4D · DeformingThingsIoU: 0.4034 -> 0.4191No clear effect
add attention mixing at inference (ablation)COM4D · CMU Panoptic point cloudsaverage Chamfer Distance (cm): 35.91 -> 7.42Improved
add attention mixing at inference (ablation)COM4D · User study (spatial correctness, temporal coherence)preference rate: 87% vs. 6.9%Improved
replace frame-wise TripoSG with COM4D temporal modelingTripoSG · DeformingThings4DF-Score: 0.5179 -> 0.8388Improved
use COM4D instead of TripoSGsingle-object 4D · Objaverse 4D subsetChamfer Distance: 0.1107 -> 0.1205Hurt
use COM4D instead of MIDI on 3D scenesMIDI · 3D-FRONTChamfer Distance: 0.1445 -> 0.0909Improved
use COM4D instead of PartCrafter on 3D scenesPartCrafter · 3D-FRONT part IoU (lower better)IoU: 0.0017 -> 0.0018Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 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