Single-image Full-body Human Relighting

Simple item page
dc.contributor.authorLagunas, Manuelen_US
dc.contributor.authorSun, Xinen_US
dc.contributor.authorYang, Jimeien_US
dc.contributor.authorVillegas, Rubenen_US
dc.contributor.authorZhang, Jianmingen_US
dc.contributor.authorShu, Zhixinen_US
dc.contributor.authorMasia, Belenen_US
dc.contributor.authorGutierrez, Diegoen_US
dc.contributor.editorBousseau, Adrien and McGuire, Morganen_US
dc.date.accessioned2021-07-12T12:13:41Z
dc.date.available2021-07-12T12:13:41Z
dc.date.issued2021
dc.description.abstractWe present a single-image data-driven method to automatically relight images with full-body humans in them. Our framework is based on a realistic scene decomposition leveraging precomputed radiance transfer (PRT) and spherical harmonics (SH) lighting. In contrast to previous work, we lift the assumptions on Lambertian materials and explicitly model diffuse and specular reflectance in our data. Moreover, we introduce an additional light-dependent residual term that accounts for errors in the PRTbased image reconstruction. We propose a new deep learning architecture, tailored to the decomposition performed in PRT, that is trained using a combination of L1, logarithmic, and rendering losses. Our model outperforms the state of the art for full-body human relighting both with synthetic images and photographs.en_US
dc.description.sectionheadersFaces and Body
dc.description.seriesinformationEurographics Symposium on Rendering - DL-only Track
dc.identifier.doi10.2312/sr.20211300
dc.identifier.isbn978-3-03868-157-1
dc.identifier.issn1727-3463
dc.identifier.pages167-177
dc.identifier.urihttps://doi.org/10.2312/sr.20211300
dc.identifier.urihttps://diglib.eg.org:443/handle/10.2312/sr20211300
dc.publisherThe Eurographics Associationen_US
dc.subjectComputing methodologies
dc.subjectRendering
dc.subjectNeural networks
dc.subjectImage
dc.subjectbased rendering
dc.titleSingle-image Full-body Human Relightingen_US
Files
Original bundle
Now showing 1 - 3 of 3
Thumbnail Image
Name:
167-177.pdf
Size:
22.99 MB
Format:
Adobe Portable Document Format
Download
No Thumbnail Available
Name:
paper1028_mm.mp4
Size:
166.75 MB
Format:
Unknown data format
Download
Thumbnail Image
Name:
paper1028_mm.pdf
Size:
18.3 MB
Format:
Adobe Portable Document Format
Download
Collections
Rendering 2021 - DL-only Track