Research map / 3D vision / 3D generation
GO-Renderer: Generative Object Rendering with 3D-aware Controllable Video Diffusion Models
Zekai Gu, Shuoxuan Feng, Yansong Wang, Hanzhuo Huang, Zhongshuo Du, Chengfeng Zhao, Chengwei Ren, Peng Wang, Yuan Liu
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 negative RoPE temporal gap g=3 for reference latents (vs g=0) (ablation) | GO-Renderer · Temporal offset ablation | PSNR: PSNR drops to 15.63 at g=0 | Improved |
| use gap g=1 instead of g=0 (ablation) | GO-Renderer · Temporal offset ablation | PSNR: g=1 significantly recovers the performance | Improved |
| use g=3 rather than other gaps in {0,1,5,10} (ablation) | GO-Renderer · Temporal offset ablation | PSNR/SSIM: g=3 achieves the optimal quantitative results | Improved |
| inject increasing spatial perturbations into target coordinate maps at inference (ablation) | GO-Renderer · Proxy robustness | rendering metrics: graceful degradation then plateau | Mixed |
| add 3D proxy coordinate guidance and multi-view references (GO-Renderer vs Wan2.2Fun) | Wan2.2Fun 5B Ref Control · Relighting rendering quality (Tab. 1) | PSNR: 14.300 -> 18.260 | Improved |
| add 3D proxy coordinate guidance (GO-Renderer vs Wan2.2Fun) | Wan2.2Fun 5B Ref Control · Relighting rendering quality (Tab. 1) | SSIM: 0.641 -> 0.684 | Improved |
| use diffusion-based GO-Renderer instead of reconstruct-then-relight UniLumos(GT) | UniLumos(GT) · Relighting rendering quality (Tab. 1) | PSNR: 12.060 -> 18.260 | Improved |
| use practical 3D reconstruction (ReconViGen) instead of GT geometry in UniLumos | UniLumos · Relighting rendering quality (Tab. 1) | SSIM: 0.585 -> 0.338 | Hurt |
| use GO-Renderer instead of AnySplat reconstruction rendering | AnySplat · Multi-view consistency (blank background) | DINO Avg: 0.191 -> 0.725 | Improved |
| use GO-Renderer instead of ReconViaGen rendering | ReconViaGen · Multi-view consistency (blank background) | DINO Avg: 0.448 -> 0.725 | Improved |
| use GO-Renderer instead of single-view Phantom | Phantom · Relighting rendering quality (Tab. 1) | Aesthetic Quality: 0.568 -> 0.573 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ReconViaGen — ReconViaGen: Towards Accurate Multi-view 3D Object Reconstruction via Generation (ICLR 2026) · better on 4 of 4 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
- Diffusion Renderer: Neural Inverse and Forward Rendering with Video Diffusion Models (CVPR 2025 · similarity 0.90)
- Generative Rendering: Controllable 4D-Guided Video Generation with 2D Diffusion Models (CVPR 2024 · similarity 0.88)
- Denoising Diffusion via Image-Based Rendering (ICLR 2024 · similarity 0.87)
- DreamComposer: Controllable 3D Object Generation via Multi-View Conditions (CVPR 2024 · similarity 0.87)
- ViVid-1-to-3: Novel View Synthesis with Video Diffusion Models (CVPR 2024 · similarity 0.87)
- Decompositional Neural Scene Reconstruction with Generative Diffusion Prior (CVPR 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.
