Research map / 3D vision / 3D generation
GeoDiffusion: A Training-Free Framework for Accurate 3D Geometric Conditioning in Image Generation
Phillip Mueller, Talip Uenlue, Sebastian Schmidt, Marcel Kollovieh, Jiajie Fan, Stephan Günnemann, Lars Mikelsons
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 changed | Where | Measured | Effect |
|---|---|---|---|
| replace GoodDrag with GeoDrag in the framework | GeoDiffusion · Geometry Guidance Dataset | mean distance (px): 11.77 -> 10.80 | Improved |
| replace GoodDrag with GeoDrag on DragBench | GeoDrag · DragBench | mean distance: 24.29 -> 20.86 | Improved |
| apply GeoDrag to non-vehicle objects (chairs, buildings) | GeoDrag vs GoodDrag · 20 non-vehicle objects | dragging accuracy: +32.7% accuracy, 39.9% faster | Improved |
| skip copy-and-paste refinement after iterative optimization (ablation) | GeoDrag · Geometry Guidance Dataset | mean distance (px): +2.73 px | Hurt |
| skip refinement after each iterative optimization step (ablation) | GeoDrag · Geometry Guidance Dataset | mean distance (px): +0.19 px | Hurt |
| add SDXL image refinement after geometric modification (ablation) | GeoDiffusion · Geometry Guidance Dataset | MD and HPSv2: MD 9.35 -> 10.8; HPSv2 +1.49 | Mixed |
| do not re-enter points dragged beyond upper threshold u into optimization (ablation) | GeoDrag · Geometry Guidance Dataset | dragging accuracy: -16% | Hurt |
| run dragging for more steps / thresholds closer to target (ablation) | GeoDrag · Geometry Guidance Dataset | accuracy vs speed | Mixed |
Closest papers by meaning
- 3DGS-Drag: Dragging Gaussians for Intuitive Point-Based 3D Editing (ICLR 2025 · similarity 0.87)
- GeoComplete: Geometry-Aware Diffusion for Reference-Driven Image Completion (NeurIPS 2025 · similarity 0.87)
- FlowDrag: 3D-aware Drag-based Image Editing with Mesh-guided Deformation Vector Flow Fields (ICML 2025 · similarity 0.86)
- Geometry Image Diffusion: Fast and Data-Efficient Text-to-3D with Image-Based Surface Representation (ICLR 2025 · similarity 0.86)
- GeoCAD: Local Geometry-Controllable CAD Generation with Large Language Models (NeurIPS 2025 · similarity 0.85)
- GeoDiffusion: Text-Prompted Geometric Control for Object Detection Data Generation (ICLR 2024 · similarity 0.85)
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.
