Research map / 3D vision / Face and shape
RenderMe-360: A Large Digital Asset Library and Benchmarks Towards High-fidelity Head Avatars
Dongwei Pan, Long Zhuo, Jingtan Piao, Huiwen Luo, Wei Cheng, Yuxin WANG, Siming Fan, Shengqi Liu, Lei Yang, Bo Dai, Ziwei Liu, Chen Change Loy and 4 more
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 |
|---|---|---|---|
| Train with random source views rather than fixed source views | Generalizable NVS methods · Unseen expressions and unseen identities | PSNR | Improved |
| Increase human-head appearance and geometry complexity | Single-ID NVS methods · Normal, deformable-accessory, and complex-accessory cases | PSNR: 24.71 -> 20.54 | Hurt |
| Use a surface representation instead of volumetric representation | Single-ID NVS · Overall | PSNR: 23.21 -> 24.67 | Improved |
| Use PointAvatar's point-based representation with an implicit representation | Novel expression synthesis · Normal and hard expressions | PSNR: 22.61 -> 24.57; 21.91 -> 25.05 | Improved |
| Use point-based representation for subtle facial motions | PointAvatar · Novel expression synthesis | PSNR | Hurt |
| Increase hair geometry complexity from short or long hair to curls | Hair-rendering methods · Static and dynamic novel-view hair rendering | PSNR | Hurt |
| Use NSFF instead of NR-NeRF for time interpolation of dynamic hair | Hair rendering · Time-Interpolation | LPIPS: 0.124 -> 0.098 | Improved |
| Use Instant-NGP instead of NeuS for static hair rendering | Hair rendering · Static Rendering, Overall | PSNR: 21.76 -> 23.75 | Improved |
| Use MVP instead of NV for dynamic hair rendering | Hair rendering · Dynamic Rendering, Overall | PSNR: 20.32 -> 22.61 | Improved |
| Replace e4e inversion with Restyle_e4e inversion | HairCLIP hair editing · Normal hairstyles | ID-score: 0.57 -> 0.58 | Improved |
| Use PTI inversion rather than e4e inversion | Text-based hair editing · Normal hairstyles | ID-score: 0.57 -> 0.88 | Improved |
| Evaluate on deformable-accessory images instead of normal hairstyles | Hair editing inversion methods · Hard cases | ID-score: 0.88 -> 0.86 | Hurt |
Closest papers by meaning
- Neural Head Avatars From Monocular RGB Videos (CVPR 2022 · similarity 0.90)
- High-Fidelity 3D Head Avatars Reconstruction through Spatially-Varying Expression Conditioned Neural Radiance Field (AAAI 2024 · similarity 0.90)
- Gaussian Head Avatar: Ultra High-fidelity Head Avatar via Dynamic Gaussians (CVPR 2024 · similarity 0.89)
- MeGA: Hybrid Mesh-Gaussian Head Avatar for High-Fidelity Rendering and Head Editing (CVPR 2025 · similarity 0.89)
- LUCAS: Layered Universal Codec Avatars (CVPR 2025 · similarity 0.89)
- Efficient 3D Implicit Head Avatar with Mesh-anchored Hash Table Blendshapes (CVPR 2024 · similarity 0.88)
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.
