Research map / 3D vision / Face and shape
AvatarMe: Realistically Renderable 3D Facial Reconstruction "In-the-Wild"
Check 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 geometry depth to the de-lighting network input (ablation) | AvatarMe · RealFaceDB de-lighting | diffuse albedo quality | Helped |
| Use texture alone instead of texture with geometry (ablation) | AvatarMe de-lighting network · RealFaceDB de-lighting | diffuse albedo quality | Hurt |
| Use VGG feature-matching loss (ablation) | AvatarMe image-translation networks · Image-to-image translation tasks | performance | Hurt |
| Use high-resolution patches for inference instead of inferring the whole face at lower resolution (ablation) | AvatarMe reflectance-map inference · In-the-wild face reconstruction | detail quality | Helped |
| Use diffuse albedo rather than the reconstructed texture containing highlights as input (ablation) | AvatarMe specular-normal network · RealFaceDB specular-normal estimation | specular-normal quality | Helped |
| Add shape normals as geometric guidance to the translation-network input (ablation) | AvatarMe image-translation networks · Reflectance-map estimation | output detail | Helped |
| Use tangent-space normals rather than object-space normals for specular-normal estimation (ablation) | AvatarMe specular-normal network · RealFaceDB specular-normal estimation | high-frequency detail | Helped |
| Use diffuse albedo rather than baked-illumination texture to infer specular albedo (ablation) | AvatarMe specular-albedo network · RealFaceDB specular-albedo estimation | specular-albedo quality | Helped |
| Use patch-based translation training instead of training on full-resolution maps (ablation) | AvatarMe image-translation networks · High-resolution reflectance estimation | training feasibility | Helped |
| Use super-resolution followed by de-lighting and reflectance-component estimation | AvatarMe · In-the-wild face reconstruction | rendering quality | Helped |
Closest papers by meaning
- High-Fidelity 3D Head Avatars Reconstruction through Spatially-Varying Expression Conditioned Neural Radiance Field (AAAI 2024 · similarity 0.88)
- CAP4D: Creating Animatable 4D Portrait Avatars with Morphable Multi-View Diffusion Models (CVPR 2025 · similarity 0.88)
- RGBAvatar: Reduced Gaussian Blendshapes for Online Modeling of Head Avatars (CVPR 2025 · similarity 0.87)
- Vid2Avatar-Pro: Authentic Avatar from Videos in the Wild via Universal Prior (CVPR 2025 · similarity 0.87)
- GAF: Gaussian Avatar Reconstruction from Monocular Videos via Multi-view Diffusion (CVPR 2025 · similarity 0.87)
- DiffHuman: Probabilistic Photorealistic 3D Reconstruction of Humans (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.