BEGIN:VCALENDAR VERSION:2.0 PRODID:Linklings LLC BEGIN:VTIMEZONE TZID:Asia/Seoul X-LIC-LOCATION:Asia/Seoul BEGIN:STANDARD TZOFFSETFROM:+0900 TZOFFSETTO:+0900 TZNAME:KST DTSTART:18871231T000000 DTSTART:19881009T020000 END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20230103T035307Z LOCATION:Auditorium\, Level 5\, West Wing DTSTART;TZID=Asia/Seoul:20221206T100000 DTEND;TZID=Asia/Seoul:20221206T120000 UID:siggraphasia_SIGGRAPH Asia 2022_sess153_papers_409@linklings.com SUMMARY:Learning-based Inverse Rendering of Complex Indoor Scenes with Dif ferentiable Monte Carlo Raytracing DESCRIPTION:Technical Papers\n\nLearning-based Inverse Rendering of Comple x Indoor Scenes with Differentiable Monte Carlo Raytracing\n\nZhu, Luan, H uo, Lin, Zhong...\n\nIndoor scenes typically exhibit complex, spatially-va rying appearance from global illumination, making the inverse rendering a challenging ill-posed problem. This work presents an end-to-end, learning- based inverse rendering framework incorporating differentiable Monte Carlo raytracing with importance sampling. The framework takes a single image a s input to jointly recover the underlying geometry, spatially-varying ligh ting, and photorealistic materials. Specifically, we introduce a physicall y-based differentiable rendering layer with screen-space ray tracing, resu lting in more realistic specular reflections that match the input photo. I n addition, we create a large-scale, photorealistic indoor scene dataset w ith significantly richer details like complex furniture and dedicated deco rations. Further, we design a novel out-of-view lighting network with unce rtainty-aware refinement leveraging hypernetwork-based neural radiance fie lds to predict lighting outside the view of the input photo. Through exten sive evaluations on common benchmark datasets, we demonstrate superior inv erse rendering quality of our method compared to state-of-the-art baseline s, enabling various applications such as complex object insertion and mate rial editing with high fidelity. Code and data will be made available at h ttps://jingsenzhu.github.io/invrend.\n\nRegistration Category: FULL ACCESS , EXPERIENCE PLUS ACCESS, EXPERIENCE ACCESS, TRADE EXHIBITOR\n\nLanguage: ENGLISH\n\nFormat: IN-PERSON URL:https://sa2022.siggraph.org/en/full-program/?id=papers_409&sess=sess15 3 END:VEVENT END:VCALENDAR