Research map / 3D vision / Face and shape
AUTO3D: Novel view synthesis through unsupervisely learned variational viewpoint and global 3D representation
Xiaofeng Liu, Tong Che, Yiqun Lu, Chao Yang, Site Li, Jane You
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 AdaIN conditioning of the pose code in the decoder (ablation) | AUTO3D · ShapeNet Chair (single input) | SSIM: 0.83 -> 0.87 | Improved |
| add AdaIN conditioning of the pose code in the decoder (ablation) | AUTO3D · ShapeNet Chair (single input) | L1 error: 12.65 -> 12.62 | No clear effect |
| add adversarial (GAN) loss (ablation) | AUTO3D · ShapeNet Chair (single input) | SSIM: 0.83 -> 0.87 | Improved |
| add tensor slides (ablation) | AUTO3D · ShapeNet Chair (single input) | SSIM: 0.85 -> 0.87 | Improved |
| add tensor slides for training speed (ablation) | AUTO3D · ShapeNet | training speed: 1.5 times | Improved |
| add spatial correlation module (SCM) with a single input (ablation) | AUTO3D · ShapeNet Chair (single input) | L1 error: 12.62 -> 12.62 | No clear effect |
| add SCM with multiple input views (ablation) | AUTO3D · ShapeNet Chair (4 input views) | L1 error: 0.112 -> 0.106 | Improved |
| add SCM with multiple input views (ablation) | AUTO3D · ShapeNet Chair (4 input views) | SSIM: 0.929 -> 0.938 | Improved |
| add spatial-aware Gaussian term to SCM (ablation) | AUTO3D SCM · ShapeNet Chair (4 input views) | L1 error: 0.110 -> 0.106 | Improved |
| add local diffusion-based complementary-aware formulation to SCM (ablation) | AUTO3D SCM · ShapeNet Chair (4 input views) | L1 error: 0.108 -> 0.106 | Improved |
| pre-train the encoder with unsupervised AUTO3D before 3D landmark training | 3D face landmark estimator · AFLW2000-3D 3D landmark estimation | mean NME (%): 12.7% -> 6.9% | Improved |
| increase the number of input views from 1 to 4 | AUTO3D · ShapeNet Chair | L1 error: 0.182 -> 0.106 | Improved |
Closest papers by meaning
- Free3D: Consistent Novel View Synthesis without 3D Representation (CVPR 2024 · similarity 0.88)
- UMAMI: Unifying Masked Autoregressive Models and Deterministic Rendering for View Synthesis (NeurIPS 2025 · similarity 0.88)
- Less Decoder is More Encoder: Geometric Representation Learning from Novel View Synthesis (NeurIPS 2026 · similarity 0.87)
- NAMVIS: Next-Scale Autoregressive Multi-View Image Synthesis (NeurIPS 2026 · similarity 0.87)
- The More You See in 2D the More You Perceive in 3D (CVPR 2024 · similarity 0.87)
- Unifying Correspondence Pose and NeRF for Generalized Pose-Free Novel View Synthesis (CVPR 2024 · similarity 0.86)
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.
