Research map / 3D vision / NeRF and radiance fields
Learning Detailed Radiance Manifolds for High-Fidelity and 3D-Consistent Portrait Synthesis From Monocular Image
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 |
|---|---|---|---|
| Replace the MLP radiance generator and use lower-resolution ray-manifold intersection calculation | GRAM · FFHQ generated images | Memory: 12G -> 3.3G | Helped |
| Replace the MLP radiance generator and use lower-resolution ray-manifold intersection calculation | GRAM · FFHQ generated images | FPS: 2 -> 14 | Helped |
| Replace the MLP radiance generator and use lower-resolution ray-manifold intersection calculation | GRAM · FFHQ generated images | FID: 15.0 -> 14.2 | Helped |
| Add learned high-resolution detail manifolds (ablation) | General encoder-based GRAM inversion · CelebA-HQ test, inversion | PSNR: 17.68 -> 21.51 | Helped |
| Pretrain efficient GRAM before training the inverter (ablation) | General inversion · CelebA-HQ test, inversion | ID: 0.472 -> 0.648 | Helped |
| Finetune the pretrained efficient GRAM (ablation) | General inversion · CelebA-HQ test, inversion | PSNR: 17.68 -> 18.00 | Helped |
| Represent the low-resolution detail voxel in camera space rather than world space (ablation) | Detail manifolds reconstructor · CelebA-HQ test, inversion | PSNR: 19.96 -> 21.51 | Helped |
| Add manifold super-resolution for high-resolution detail prediction (ablation) | Detail manifolds reconstructor · CelebA-HQ test, inversion | LPIPS: 0.165 -> 0.127 | Helped |
| Add novel-view regularization (ablation) | Detail manifolds reconstructor · CelebA-HQ test, novel views | FID_nv: 36.49 -> 36.02 | Helped |
| Use a normal-aware mask with novel-view regularization (ablation) | Detail manifolds reconstructor · CelebA-HQ test, novel views | FID_nv: 43.25 -> 36.02 | Helped |
| Add depth regularization to the detail manifolds (ablation) | Detail manifolds reconstructor · CelebA-HQ test, novel views | FID_nv: 35.60 -> 36.02 | Mixed |
Closest papers by meaning
- GRAM: Generative Radiance Manifolds for 3D-Aware Image Generation (CVPR 2022 · similarity 0.91)
- GRAF: Generative Radiance Fields for 3D-Aware Image Synthesis (NeurIPS 2020 · similarity 0.90)
- SinGRAF: Learning a 3D Generative Radiance Field for a Single Scene (CVPR 2023 · similarity 0.89)
- Semantic 3D-Aware Portrait Synthesis and Manipulation Based on Compositional Neural Radiance Field (AAAI 2023 · similarity 0.88)
- Multi-View Consistent Generative Adversarial Networks for 3D-Aware Image Synthesis (CVPR 2022 · similarity 0.88)
- In-N-Out: Faithful 3D GAN Inversion with Volumetric Decomposition for Face Editing (CVPR 2024 · 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.