Open tkuri opened 2 years ago
多視点RGB画像から物体のSVBRDFを推定する。従来手法では間接照明はモデル化しない事が多く特にアルベドに弊害が出る。そこで間接照明を照明とMaterialから直接推定するのではなく入力画像から学習したNeural Radiance Fieldから簡便に導出することができることを示した。 https://openaccess.thecvf.com/content/CVPR2022/html/Zhang_Modeling_Indirect_Illumination_for_Inverse_Rendering_CVPR_2022_paper.html
未確認。
論文概要
多視点RGB画像から物体のSVBRDFを推定する。従来手法では間接照明はモデル化しない事が多く特にアルベドに弊害が出る。そこで間接照明を照明とMaterialから直接推定するのではなく入力画像から学習したNeural Radiance Fieldから簡便に導出することができることを示した。 https://openaccess.thecvf.com/content/CVPR2022/html/Zhang_Modeling_Indirect_Illumination_for_Inverse_Rendering_CVPR_2022_paper.html
Code
未確認。