Research map / 3D vision / NeRF and radiance fields
Sharf: Shape-conditioned Radiance Fields from a Single View
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 an explicit shape scaffold rather than using a single entangled appearance code (ablation) | ShaRF · ShapeNet-SRN Chairs | PSNR: 22.12 -> 23.37 | Helped |
| Add an explicit shape scaffold rather than using a single entangled appearance code (ablation) | ShaRF · ShapeNet-Realistic Chairs | PSNR: 21.86 -> 21.68 (ShapeFromNR); 21.86 -> 22.68 (ShapeFromMask) | Mixed |
| Estimate the scaffold from an input mask instead of image reconstruction (ablation) | ShaRF · ShapeNet-SRN Chairs | PSNR: 23.37 -> 22.94 | Hurt |
| Estimate the scaffold from an input mask instead of image reconstruction (ablation) | ShaRF · ShapeNet-Realistic Chairs | PSNR: 21.68 -> 22.68 | Helped |
| Replace the predicted scaffold with ground-truth shape (ablation) | ShaRF · ShapeNet-SRN Chairs | PSNR: 23.37 -> 25.59 | Helped |
| Replace the predicted scaffold with ground-truth shape (ablation) | ShaRF · ShapeNet-Realistic Chairs | PSNR: 22.68 -> 24.84 | Helped |
| Fine-tune the appearance network together with its latent code (ablation) | ShaRF ShapeFromNR · ShapeNet-SRN Chairs | PSNR: 23.37 -> 23.31 | No clear effect |
| Fine-tune the appearance network together with its latent code (ablation) | ShaRF · ShapeNet-Realistic Chairs | PSNR: 21.86 -> 22.91 (Conditional NeRF); 21.68 -> 22.07 (ShapeFromNR); 22.68 -> 23.26 (ShapeFromMask) | Helped |
| Add mirrored-input symmetry during test-time optimization (ablation) | ShaRF ShapeFromMask · ShapeNet-Realistic | PSNR: 23.26 -> 23.51 | Helped |
| Use the shape scaffold rather than an unconditioned single-code field (ablation) | ShaRF ShapeFromMask · ShapeNet-Realistic Chairs | LPIPS: 0.115 -> 0.109 | Helped |
| Use the learned shape scaffold for single-view shape reconstruction | ShaRF ShapeFromMask · ShapeNet Chair shape reconstruction | IoU: 0.50 vs 0.49 | No clear effect |
| Use the learned shape scaffold for single-view shape reconstruction | ShaRF ShapeFromMask · ShapeNet Car shape reconstruction | IoU: 0.74 -> 0.77 | Helped |
Closest papers by meaning
- AutoRF: Learning 3D Object Radiance Fields From Single View Observations (CVPR 2022 · similarity 0.91)
- pixelNeRF: Neural Radiance Fields From One or Few Images (CVPR 2021 · similarity 0.91)
- GRAF: Generative Radiance Fields for 3D-Aware Image Synthesis (NeurIPS 2020 · similarity 0.91)
- Mask-Based Modeling for Neural Radiance Fields (ICLR 2024 · similarity 0.89)
- NeRFPrior: Learning Neural Radiance Field as a Prior for Indoor Scene Reconstruction (CVPR 2025 · similarity 0.89)
- Learning Robust Generalizable Radiance Field with Visibility and Feature Augmented Point Representation (ICLR 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.