AI Paper Atlas

Research map / 3D vision / Gaussian splatting

GIR: 3D Gaussian Inverse Rendering for Relightable Scene Factorization

Yahao Shi, Yanmin Wu, Chenming Wu, Xing Liu, Chen Zhao, Haocheng Feng, Jian Zhang, Bin Zhou, Errui Ding, Jingdong Wang

TPAMI 2025 · 3D vision › Gaussian splatting · approach: Sparse-view Gaussian splatting for novel views · 82 citations (OpenAlex)

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 changedWhereMeasuredEffect
add FCN-parameterized environment map (ablation)GIR · Shiny Blender NVSPSNR: 32.283 -> 33.241Improved
add geometry regularization on top of FCN (ablation)GIR · Shiny Blender NVSPSNR: 33.241 -> 33.432Improved
add material regularization on top of FCN (ablation)GIR · Shiny Blender NVSPSNR: 33.241 -> 33.377Improved
add both geometry and material regularization (ablation)GIR · Shiny Blender NVSPSNR: 33.241 -> 33.759Improved
add voxel-based indirect illumination estimation (ablation)GIR · Shiny Blender NVSPSNR: 33.759 -> 34.513Improved
add voxel-based indirect illumination estimation (ablation)GIR · Shiny Blender NVSLPIPS: 0.059 -> 0.057Improved
add directional masking scheme (ablation)GIR · Normal estimation (ball)normal accuracy (qualitative)Improved
compare GIR with TensoIR on relighting SSIMGIR · TensoIR dataset relightingSSIM: 0.945 -> 0.943No clear effect
GIR versus TensoIR relightingGIR · TensoIR dataset relightingLPIPS: 0.080 -> 0.059Improved
GIR versus 3DGS rendering speedGIR · Rendering speed (350K Gaussians, V100)FPS: 120.13 -> 92.58Hurt
GIR versus 3DGS rendering speed in UE5GIR · UE5 plugin rendering speedFPS: 85.57 -> 78.75Hurt
GIR versus GS Shader on NVS PSNRGIR · TensoIR dataset NVSPSNR: 37.573 -> 37.626Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy