Training-Free Multi-View Extension of IC-Light for Textual Position-Aware Scene Relighting

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Ye, Jiangnan, Zhuang, Jiedong, Mu, Lianrui, Zheng, Wenjie, Hu, Jiaqi, Zou, Xingze, Wang, Jing, Hu, Haoji
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866915623296040960
author Ye, Jiangnan
Zhuang, Jiedong
Mu, Lianrui
Zheng, Wenjie
Hu, Jiaqi
Zou, Xingze
Wang, Jing
Hu, Haoji
author_facet Ye, Jiangnan
Zhuang, Jiedong
Mu, Lianrui
Zheng, Wenjie
Hu, Jiaqi
Zou, Xingze
Wang, Jing
Hu, Haoji
contents We introduce GS-Light, an efficient, textual position-aware pipeline for text-guided relighting of 3D scenes represented via Gaussian Splatting (3DGS). GS-Light implements a training-free extension of a single-input diffusion model to handle multi-view inputs. Given a user prompt that may specify lighting direction, color, intensity, or reference objects, we employ a large vision-language model (LVLM) to parse the prompt into lighting priors. Using off-the-shelf estimators for geometry and semantics (depth, surface normals, and semantic segmentation), we fuse these lighting priors with view-geometry constraints to compute illumination maps and generate initial latent codes for each view. These meticulously derived init latents guide the diffusion model to generate relighting outputs that more accurately reflect user expectations, especially in terms of lighting direction. By feeding multi-view rendered images, along with the init latents, into our multi-view relighting model, we produce high-fidelity, artistically relit images. Finally, we fine-tune the 3DGS scene with the relit appearance to obtain a fully relit 3D scene. We evaluate GS-Light on both indoor and outdoor scenes, comparing it to state-of-the-art baselines including per-view relighting, video relighting, and scene editing methods. Using quantitative metrics (multi-view consistency, imaging quality, aesthetic score, semantic similarity, etc.) and qualitative assessment (user studies), GS-Light demonstrates consistent improvements over baselines. Code and assets will be made available upon publication.
format Preprint
id arxiv_https___arxiv_org_abs_2511_13684
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Training-Free Multi-View Extension of IC-Light for Textual Position-Aware Scene Relighting
Ye, Jiangnan
Zhuang, Jiedong
Mu, Lianrui
Zheng, Wenjie
Hu, Jiaqi
Zou, Xingze
Wang, Jing
Hu, Haoji
Computer Vision and Pattern Recognition
Machine Learning
We introduce GS-Light, an efficient, textual position-aware pipeline for text-guided relighting of 3D scenes represented via Gaussian Splatting (3DGS). GS-Light implements a training-free extension of a single-input diffusion model to handle multi-view inputs. Given a user prompt that may specify lighting direction, color, intensity, or reference objects, we employ a large vision-language model (LVLM) to parse the prompt into lighting priors. Using off-the-shelf estimators for geometry and semantics (depth, surface normals, and semantic segmentation), we fuse these lighting priors with view-geometry constraints to compute illumination maps and generate initial latent codes for each view. These meticulously derived init latents guide the diffusion model to generate relighting outputs that more accurately reflect user expectations, especially in terms of lighting direction. By feeding multi-view rendered images, along with the init latents, into our multi-view relighting model, we produce high-fidelity, artistically relit images. Finally, we fine-tune the 3DGS scene with the relit appearance to obtain a fully relit 3D scene. We evaluate GS-Light on both indoor and outdoor scenes, comparing it to state-of-the-art baselines including per-view relighting, video relighting, and scene editing methods. Using quantitative metrics (multi-view consistency, imaging quality, aesthetic score, semantic similarity, etc.) and qualitative assessment (user studies), GS-Light demonstrates consistent improvements over baselines. Code and assets will be made available upon publication.
title Training-Free Multi-View Extension of IC-Light for Textual Position-Aware Scene Relighting
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2511.13684