Research map / 3D vision / Face and shape
MOST-GAN: 3D Morphable StyleGAN for Disentangled Face Image Manipulation
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 |
|---|---|---|---|
| Use explicit 3D pose control and 3D-guided hair warping instead of PIE latent-space editing | MOST-GAN versus PIE · Head pose editing across -90° to 90° | mean absolute pose error (degrees): MOST-GAN 13.2, 6.0, 3.5, 4.5, 4.7, 4.8, 2.3, 2.6, 5.0, 6.0, 3.9, 2.7, 6.7; PIE 61.7, 38.0, 22.1, 15.8, 10.8, 7.6, 3.5, 6 | Helped |
| Add the refiner (ablation) | MOST-GAN · MOST-GAN test reconstructions | FR score: 0.66 ± 0.10 -> 0.68 ± 0.10 | Helped |
| Add the refiner (ablation) | MOST-GAN · MOST-GAN test reconstructions | SSIM: 0.72 ± 0.09 -> 0.74 ± 0.08 | Helped |
| Add the refiner (ablation) | MOST-GAN · MOST-GAN test reconstructions | PSNR: 22.24 ± 2.63 -> 23.20 ± 2.47 | Helped |
| Add the refiner (ablation) | MOST-GAN · Masked face reconstruction | FR score: 0.69 ± 0.10 -> 0.71 ± 0.10 | Helped |
| Add the refiner (ablation) | MOST-GAN · Masked face reconstruction | SSIM: 0.87 ± 0.04 -> 0.88 ± 0.04 | Helped |
| Add the refiner (ablation) | MOST-GAN · Masked face reconstruction | PSNR: 25.08 ± 2.92 -> 26.17 ± 2.71 | Helped |
| Pretrain on synthetic faces before real-image training (ablation) | MOST-GAN · Real face reconstruction | reconstruction stability | Helped |
| Remove the albedo and lighting regularization losses (ablation) | MOST-GAN real-data training · Real face reconstruction | albedo-lighting disentanglement | Hurt |
| Remove any of the other real-data loss functions (ablation) | MOST-GAN real-data training · Real face reconstruction | reconstruction quality | Hurt |
Closest papers by meaning
- Interpreting the Latent Space of GANs for Semantic Face Editing (CVPR 2020 · similarity 0.90)
- Mask-Guided Portrait Editing With Conditional GANs (CVPR 2019 · similarity 0.89)
- In-N-Out: Faithful 3D GAN Inversion with Volumetric Decomposition for Face Editing (CVPR 2024 · similarity 0.87)
- CGS-GAN: 3D Consistent Gaussian Splatting GANs for High Resolution Human Head Synthesis (NeurIPS 2025 · similarity 0.87)
- Dual Mapping of 2D StyleGAN for 3D-Aware Image Generation and Manipulation (Student Abstract) (AAAI 2024 · similarity 0.87)
- The GAN is dead; long live the GAN! A Modern GAN Baseline (NeurIPS 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.