AI Paper Atlas

Research map / 3D vision / 3D generation

SpatialDiff: 3D-Aware Object Movement via Implicit Spatial Modeling

Zheng Liu, Zijian He, Huiguo He, Weizhi Zhong, Yejun Tang, Huan Yang, Kun Gai, Guanbin Li

CVPR 2026 · 3D vision › 3D generation · approach: Open-vocabulary 3D scene understanding · 0 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 LoRA fine-tuning of DiT attention on the movement dataset (ablation)Baseline editing model · SpatialBench ablationGPT-SC: 0.236 -> 0.398Improved
add LoRA fine-tuning of DiT attention on the movement dataset (ablation)Baseline editing model · SpatialBench ablationGPT-PQ: 0.831 -> 0.795Hurt
add Implicit 3D Spatial Modeling (3D tokens via foundation model) (ablation)Model A · SpatialBench ablationGPT-SC: 0.398 -> 0.518Improved
add Implicit 3D Spatial Modeling (ablation)Model A · SpatialBench ablationGPT-PQ: 0.795 -> 0.743Hurt
add explicit pixel-space depth supervision (ablation)Model B · SpatialBench ablationGPT-O: 0.620 -> 0.673Improved
use latent depth supervision instead of explicit pixel-space depth supervision (ablation)Model C · SpatialBench ablationGPT-SC: 0.566 -> 0.804Improved
use latent depth supervision instead of explicit pixel-space depth supervision (ablation)Model C · SpatialBench ablationGPT-PQ: 0.801 -> 0.871Improved
add Implicit 3D Spatial Modeling (VGGT features) over learnable queries (ablation)SpatialDiff with LDS · SpatialBench ablationGPT-SC: 0.656 -> 0.804Improved
add latent depth supervision to FT+ISM (ablation)SpatialDiff · SpatialBench ablationGPT-O: 0.620 -> 0.837Improved
use SpatialDiff instead of Qwen-Image-EditingImage editing · SpatialBenchGPT-SC: 0.666 -> 0.803Improved
use SpatialDiff instead of Nano Banana 2Image editing · SpatialBenchQwen-SC: 0.594 -> 0.778Improved
use SpatialDiff instead of Nano Banana 2Image editing · SpatialBenchGPT-PQ: 0.892 -> 0.886No 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