Research map / 3D vision / 3D generation
SViM3D: Stable Video Material Diffusion for Single Image 3D Generation
Andreas Engelhardt, Mark Boss, Vikram Voleti, Chun-Han Yao, Hendrik P. A. Lensch, Varun Jampani
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 homography correction in 3D reconstruction (ablation) | SViM3D 3D pipeline · Poly Haven subset | PSNR: 13.7 -> 22.4 | Improved |
| add fast differentiable environment-based lighting (ablation) | SViM3D 3D pipeline · Poly Haven subset | PSNR: 15.0 -> 22.4 | Improved |
| add view-dependent masking (ablation) | SViM3D 3D pipeline · Poly Haven subset | PSNR: 17.8 -> 22.4 | Improved |
| jointly generate PBR materials and normals with RGB | SViM3D · Poly Haven multi-view RGB NVS | PSNR: SV3D 18.41 -> SViM3D 19.57 | Improved |
| predict PBR materials in a multi-view consistent model versus running single-image material estimators on SV3D views | SViM3D · Poly Haven multi-view basecolor | FID: 17.22 -> 9.42 | Improved |
| predict PBR materials in a multi-view model versus SV3D+SM | SViM3D · Poly Haven multi-view basecolor | CMMD: 0.96 -> 1.16 | Hurt |
| use SViM3D instead of SF3D on real-world Stanford Orb | SViM3D · Stanford Orb | relighting PSNR: 20.1 -> 21.86 | Improved |
| freeze temporal attention blocks in first training phase then fine-tune all | SViM3D training · training schedule | stability and temporal knowledge | Improved |
| relighting via 2.5D rendering with generated PBR instead of raw RGB output | SViM3D · Poly Haven multi-view NVS | FID: PSNR 19.57 -> 19.99; FID 6.93 -> 15.15 | Mixed |
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.
- SF3D — SF3D: Stable Fast 3D Mesh Reconstruction with UV-unwrapping and Illumination Disentanglement (CVPR 2025) · better on 2 of 2 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
- SV4D: Dynamic 3D Content Generation with Multi-Frame and Multi-View Consistency (ICLR 2025 · similarity 0.90)
- Diffusion Renderer: Neural Inverse and Forward Rendering with Video Diffusion Models (CVPR 2025 · similarity 0.89)
- Material Palette: Extraction of Materials from a Single Image (CVPR 2024 · similarity 0.89)
- Material Anything: Generating Materials for Any 3D Object via Diffusion (CVPR 2025 · similarity 0.89)
- DreamCraft3D: Hierarchical 3D Generation with Bootstrapped Diffusion Prior (ICLR 2024 · similarity 0.89)
- One-2-3-45++: Fast Single Image to 3D Objects with Consistent Multi-View Generation and 3D Diffusion (CVPR 2024 · similarity 0.89)
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.
