Research map / 3D vision / Face and shape
DELIFFAS: Deformable Light Fields for Fast Avatar Synthesis
YoungJoong Kwon, Lingjie Liu, Henry Fuchs, Marc Habermann, Christian Theobalt
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 |
|---|---|---|---|
| replace single-surface with two-surface parameterization (L1 only) (ablation) | DELIFFAS · DynaCap D2 novel view synthesis | PSNR: 31.34 -> 33.27 | Improved |
| replace single-surface with two-surface parameterization (L1 only) (ablation) | DELIFFAS · DynaCap D2 novel view synthesis | LPIPS: 22.74 -> 18.51 | Improved |
| condition single-surface MLP on viewing direction (ablation) | single-surface · DynaCap D2 novel view synthesis | PSNR: 31.34 -> 32.19 | Improved |
| add perceptual loss to L1 (ablation) | two-surface DELIFFAS · DynaCap D2 novel view synthesis | LPIPS: 18.51 -> 12.85 | Improved |
| add perceptual loss to L1 (ablation) | two-surface DELIFFAS · DynaCap D2 novel view synthesis | FID: 23.58 -> 8.69 | Improved |
| add perceptual loss to L1 (ablation) | two-surface DELIFFAS · DynaCap D2 novel view synthesis | PSNR: 33.27 -> 33.30 | No clear effect |
| use smoothed coarse template mesh (ablation) | two-surface DELIFFAS · DynaCap D2 novel view synthesis | PSNR: 33.27 -> 32.03 | Hurt |
| two-surface instead of single-surface with coarse mesh (ablation) | DELIFFAS · DynaCap D2 novel view synthesis | PSNR: 29.63 -> 32.03 | Improved |
| image-level instead of patch-level perceptual supervision (ablation) | DELIFFAS · DynaCap D2 novel view synthesis | LPIPS: 16.54 -> 12.85 | Improved |
| image-level instead of patch-level perceptual supervision (ablation) | DELIFFAS · DynaCap D2 novel view synthesis | FID: 18.96 -> 8.69 | Improved |
| use single-surface with two U-Nets instead of two-surface (ablation) | single-surface · DynaCap D2 novel view synthesis | PSNR: 33.27 -> 31.47 | Hurt |
| add a second U-Net to single-surface (L1) (ablation) | single-surface · DynaCap D2 novel view synthesis | LPIPS: 22.74 -> 15.43 | Improved |
Closest papers by meaning
- 3DGS-Avatar: Animatable Avatars via Deformable 3D Gaussian Splatting (CVPR 2024 · similarity 0.88)
- Human Gaussian Splatting: Real-time Rendering of Animatable Avatars (CVPR 2024 · similarity 0.86)
- Relightable and Animatable Neural Avatars from Videos (AAAI 2024 · similarity 0.86)
- SplattingAvatar: Realistic Real-Time Human Avatars with Mesh-Embedded Gaussian Splatting (CVPR 2024 · similarity 0.86)
- Human-3Diffusion: Realistic Avatar Creation via Explicit 3D Consistent Diffusion Models (NeurIPS 2024 · similarity 0.86)
- ASH: Animatable Gaussian Splats for Efficient and Photoreal Human Rendering (CVPR 2024 · similarity 0.85)
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.
