AI Paper Atlas

Research map / 3D vision / Depth and stereo

Physically-Based Editing of Indoor Scene Lighting from a Single Image

Zhengqin Li, Jia Shi, Sai Bi, Rui Zhu, Kalyan Sunkavalli, Miloš Hašan, Zexiang Xu, Ravi Ramamoorthi, Manmohan Chandraker

ECCV 2022 · 3D vision › Depth and stereo · approach: Neural light field and illumination imaging · 54 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 network inpainting and denoising to Monte Carlo ray-traced shadows (ablation)Depth-based shadow rendering · Shadow rendering errorL2: 0.011 -> 0.005Improved
use RMSE Chamfer distance instead of standard L2 Chamfer for geometry loss (ablation)Light source prediction · OpenRoomstraining stabilityImproved
add L1 area loss on light sources (ablation)Light source prediction · OpenRoomslight size / shadow sharpnessImproved
use scale-invariant gradient loss instead of simple L2 for shadow network (ablation)Shadow network · OpenRoomsartifactsImproved
use log L2 loss in addition to L1/rendering loss for intensities (ablation)Light source prediction · OpenRoomshighlightsHurt
use new visible lamp representation (reflected surface plus boundary) instead of 3D bounding box (ablation)Visible lamp model · real imageshighlight artifactsImproved
add multiple importance sampling for window direct shading (ablation)Differentiable Monte Carlo renderer · window direct shadingsample efficiencyImproved
use predicted DPT depth instead of ground-truth depthLight source prediction · OpenRoomsL1 direct shading error (visible lamp): 0.317 -> 0.557 (visible lamp); 0.312 -> 0.328 (invisible window)Mixed
add rendering-loss-based refinement of light parameters (ablation)Light source prediction · real imagesrendering fidelityImproved
use optimized window parameters (sun, sky, ground) as supervision (ablation)Window prediction · OpenRoomssharpnessImproved

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