Research map / 3D vision / 3D generation
Bridging Implicit and Explicit Geometric Transformation for Single-Image View Synthesis
Byeongjun Park, Hyojun Go, Changick Kim
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 transformation similarity loss L_ts (ablation) | ViewNet · RealEstate10K (large view change) | FID: 47.74 -> 35.26 | Improved |
| add gradient detach to L_ts (ablation) | ViewNet · RealEstate10K (large view change) | FID: 47.93 -> 35.26 | Improved |
| add L_ts,in to L_ts,out (ablation) | ViewNet · RealEstate10K (large view change) | FID: 40.72 -> 35.26 | Improved |
| add L_ts,out to L_ts,in (ablation) | ViewNet · RealEstate10K (large view change) | FID: 40.17 -> 35.26 | Improved |
| add global set attention to local set attention (ablation) | ViewNet encoder · RealEstate10K (large view change) | FID: 37.63 -> 35.26 | Improved |
| add local set attention to global set attention (ablation) | ViewNet encoder · RealEstate10K (small view change) | FID: 32.80 -> 32.42 | Improved |
| replace out-of-view mask input with L_ts (ablation) | ViewNet · RealEstate10K (large view change) | FID: 46.43 -> 35.26 | Improved |
| replace DPT depth with self-supervised DepthNet (ablation) | ViewNet · RealEstate10K (large view change) | FID: 42.49 -> 35.26 | Improved |
| add the proposed adversarial loss to SynSin (ablation) | SynSin · RealEstate10K (large view change) | FID: 61.67 -> 54.27 | Improved |
| add the proposed adversarial loss to SynSin (ablation) | SynSin · RealEstate10K (large view change) | PSNR-vis: 14.72 -> 14.51 | Hurt |
| set lambda_in=10 instead of 1 (ablation) | L_ts weights · RealEstate10K (small view change) | FID: 32.42 -> 37.46 | Hurt |
| set lambda_out=0.1 instead of 1 (ablation) | L_ts weights · RealEstate10K (medium view change) | FID: 33.04 -> 37.51 | Hurt |
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.
- SynSin — SynSin: End-to-End View Synthesis From a Single Image (CVPR 2020) · better on 9 of 9 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
- UMAMI: Unifying Masked Autoregressive Models and Deterministic Rendering for View Synthesis (NeurIPS 2025 · similarity 0.89)
- Novel View Synthesis with Pixel-Space Diffusion Models (CVPR 2025 · similarity 0.88)
- LVSM: A Large View Synthesis Model with Minimal 3D Inductive Bias (ICLR 2025 · similarity 0.88)
- NAMVIS: Next-Scale Autoregressive Multi-View Image Synthesis (NeurIPS 2026 · similarity 0.88)
- MVSplat360: Feed-Forward 360 Scene Synthesis from Sparse Views (NeurIPS 2024 · similarity 0.87)
- Novel View Synthesis from A Few Glimpses via Test-Time Natural Video Completion (NeurIPS 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.
