Research map / 3D vision / 3D generation
Novel View Synthesis on Unpaired Data by Conditional Deformable Variational Auto-Encoder
Mingyu Yin, Li Sun, Qingli Li
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 conditional deformable module (CDM) (ablation) | cVAE-GAN baseline · MultiPIE (unpaired training) | FID: 44.84 -> 26.79 | Improved |
| add conditional deformable module (CDM) (ablation) | cVAE-GAN baseline · 3D chair (unpaired training) | FID: 104.78 -> 88.23 | Improved |
| add DFNM after CDM (ablation) | CDM model · MultiPIE (unpaired training) | L1 error: 23.43 -> 21.53 | Improved |
| add DFNM after CDM (ablation) | CDM model · MultiPIE (unpaired training) | FID: 26.79 -> 23.59 | Improved |
| add DFNM after CDM (ablation) | CDM model · 3D chair (unpaired training) | FID: 88.23 -> 93.11 | Hurt |
| add DFNM after CDM (ablation) | CDM model · 3D chair (unpaired training) | L1 error: 7.88 -> 6.68 | Improved |
| add domain adversarial classifier (DAC) loss on the code (ablation) | CDM+DFNM model · 3D chair (unpaired training) | L1 error: 6.68 -> 6.37 | Improved |
| add domain adversarial classifier (DAC) loss on the code (ablation) | CDM+DFNM model · MultiPIE (unpaired training) | FID: 23.59 -> 23.95 | No clear effect |
| use separate MLPs for x and y directions in CDM (ablation) | full model · MultiPIE (unpaired training) | L1 error: 24.48 -> 21.90 | Improved |
| use separate MLPs for x and y directions in CDM (ablation) | full model · 3D chair (unpaired training) | FID: 90.31 -> 86.34 | Improved |
| separate MLPs in encoder and decoder (ablation) | full model · MultiPIE (unpaired training) | FID: 28.40 -> 23.95 | Improved |
| separate MLPs in encoder and decoder (ablation) | full model · 3D chair (unpaired training) | L1 error: 6.94 -> 6.37 | Improved |
Closest papers by meaning
- AUTO3D: Novel view synthesis through unsupervisely learned variational viewpoint and global 3D representation (ECCV 2020 · similarity 0.85)
- LVSM: A Large View Synthesis Model with Minimal 3D Inductive Bias (ICLR 2025 · similarity 0.85)
- G-NeRF: Geometry-enhanced Novel View Synthesis from Single-View Images (CVPR 2024 · similarity 0.83)
- NVComposer: Boosting Generative Novel View Synthesis with Multiple Sparse and Unposed Images (CVPR 2025 · similarity 0.83)
- Novel View Synthesis with View-Dependent Effects from a Single Image (CVPR 2024 · similarity 0.83)
- Novel View Synthesis with Pixel-Space Diffusion Models (CVPR 2025 · similarity 0.83)
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.
