I2-SDF: Intrinsic Indoor Scene Reconstruction and Editing via Raytracing in Neural SDFs

Jingsen Zhu1, Yuchi Huo2,1, Qi Ye3,6, Fujun Luan4, Jifan Li1, Dianbing Xi1, Lisha Wang1, Rui Tang5, Wei Hua2, Hujun Bao1, Rui Wang1
1State Key Lab of CAD&CG, Zhejiang University, 2Zhejiang Lab, 3Zhejiang University, 4Adobe Research, 5KooLab, Manycore 6Key Lab of CS&AUS of Zhejiang Province

CVPR 2023

MonoSDF Ours
Input Relight

TL;DR: I2-SDF resolves challenges in reconstructing small objects inside an indoor scene by a novel bubble loss (Left), and leverages intrinsic decomposition and raytracing to enable photorealistic scene editing and relighting (Right).


In this work, we present I2-SDF, a new method for intrinsic indoor scene reconstruction and editing using differentiable Monte Carlo raytracing on neural signed distance fields (SDFs). Our holistic neural SDF-based framework jointly recovers the underlying shapes, incident radiance and materials from multi-view images. We introduce a novel bubble loss for fine-grained small objects and error-guided adaptive sampling scheme to largely improve the reconstruction quality on large-scale indoor scenes. Further, we propose to decompose the neural radiance field into spatially-varying material of the scene as a neural field through surface-based, differentiable Monte Carlo raytracing and emitter semantic segmentations, which enables physically based and photorealistic scene relighting and editing applications. Through a number of qualitative and quantitative experiments, we demonstrate the superior quality of our method on indoor scene reconstruction, novel view synthesis, and scene editing compared to state-of-the-art baselines.

Geometry Reconstruction

We analyze the vanishing gradient problem of SDF-based volume rendering techniques, which results in failures of existing methods to reconstruct off-surface small objects. We propose a novel bubble loss to tackle this.

Novel View Synthesis

Our method outperforms previous methods in novel view synthesis of indoor scenes. Previous methods suffers from either reconstruction failure or overfitting.

Editing and Relighting

We disentangle material (physically-based BRDF) and lighting (direct emission and indirect illumination) to enable photorealistic scene editing and relighting using Monte-Carlo raytracing.


    title = {I$^2$-SDF: Intrinsic Indoor Scene Reconstruction and Editing via Raytracing in Neural SDFs},
    author = {Jingsen Zhu and Yuchi Huo and Qi Ye and Fujun Luan and Jifan Li and Dianbing Xi and Lisha Wang and Rui Tang and Wei Hua and Hujun Bao and Rui Wang},
    booktitle = {CVPR},
    year = {2023}