Research map / 3D vision / 3D generation
CTRL&SHIFT: High-quality Geometry-Aware Object Manipulation in Visual Generation
Penghui Ruan, Bojia Zi, Xianbiao Qi, Youze Huang, Rong Xiao, Pichao WANG, Jiannong Cao, Yuhui Shi
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 |
|---|---|---|---|
| add Stage 1 object prior and pose pretraining (ablation) | Ctrl&Shift · GeoEditBench | Pose MAPE: 32.50% -> 17.70% | Improved |
| add Stage 1 object prior and pose pretraining (ablation) | Ctrl&Shift · GeoEditBench | Obj IoU: 0.68 -> 0.83 | Improved |
| add Stage 2 background preservation fine-tuning (ablation) | Ctrl&Shift · GeoEditBench | PSNR: 24.83 -> 28.71 | Improved |
| add Stage 2 background preservation fine-tuning (ablation) | Ctrl&Shift · GeoEditBench | DreamSim: 0.195 -> 0.112 | Improved |
| add auxiliary task Aux 1 (ablation) | Ctrl&Shift · GeoEditBench | CLIP-Score: 86.32 -> 90.44 | Improved |
| add auxiliary task Aux 1 (ablation) | Ctrl&Shift · GeoEditBench | Pose MAPE: 23.80% -> 17.70% | Improved |
| add auxiliary task Aux 2 (ablation) | Ctrl&Shift · GeoEditBench | Obj IoU: 0.65 -> 0.83 | Improved |
| add auxiliary task Aux 2 (ablation) | Ctrl&Shift · GeoEditBench | CLIP-Score: 83.75 -> 90.44 | Improved |
| use Ctrl&Shift instead of ObjectMover | object manipulation · ObjectMover-A (zero-shot) | PSNR: 25.27 -> 28.69 | Improved |
| use Ctrl&Shift instead of Nano-Banana | object manipulation · GeoEditBench | Pose MAPE: 24.36% -> 17.70% | Improved |
| use Ctrl&Shift instead of Nano-Banana | object manipulation · GeoEditBench | PSNR: 26.38 -> 28.71 | Improved |
Closest papers by meaning
- Thinking in Boxes: 3D Editing in Real Images Made Easy (NeurIPS 2026 · similarity 0.88)
- AffordDP: Generalizable Diffusion Policy with Transferable Affordance (CVPR 2025 · similarity 0.87)
- EC-Diffuser: Multi-Object Manipulation via Entity-Centric Behavior Generation (ICLR 2025 · similarity 0.87)
- Image Sculpting: Precise Object Editing with 3D Geometry Control (CVPR 2024 · similarity 0.87)
- AC3D: Analyzing and Improving 3D Camera Control in Video Diffusion Transformers (CVPR 2025 · similarity 0.87)
- VD3D: Taming Large Video Diffusion Transformers for 3D Camera Control (ICLR 2025 · similarity 0.87)
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.
