Our CVPR'19 paper about neural free-viewpoint rendering of human avatars without reconstructing geometry is on arXiv! Generator predicts UV mapping (not RGB), and texture is learned per-avatar. 🌐 ▶️ 📝