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_157@linklings.com SUMMARY:VToonify: Controllable High-Resolution Portrait Video Style Transf er DESCRIPTION:Technical Papers\n\nVToonify: Controllable High-Resolution Por trait Video Style Transfer\n\nYang, Jiang, Liu, Loy\n\nGenerating high-qua lity artistic portrait videos is an important and desirable task in comput er graphics and vision. Although a series of successful portrait image too nification models built upon the powerful StyleGAN have been proposed, the se image-oriented methods have obvious limitations when applied to videos, such as the fixed frame size, the requirement of face alignment, missing non-facial details and temporal inconsistency. In this work, we investigat e the challenging controllable high-resolution portrait video style transf er by introducing a novel VToonify framework. Specifically, VToonify lever ages the mid- and high-resolution layers of StyleGAN to render high-qualit y artistic portraits based on the multi-scale content features extracted b y an encoder to better preserve the frame details. The resulting fully con volutional architecture accepts non-aligned faces in videos of variable si ze as input, contributing to complete face regions with natural motions in the output. Our framework is compatible with existing StyleGAN-based imag e toonification models to extend them to video toonification, and inherits appealing features of these models for flexible style control on color an d intensity. This work presents two instantiations of VToonify built upon Toonify and DualStyleGAN for collection-based and exemplar-based portrait video style transfer, respectively. Extensive experimental results demonst rate the effectiveness of our proposed VToonify framework over existing me thods in generating high-quality and temporally-coherent artistic portrait videos with flexible style controls.\n\nRegistration Category: FULL ACCES S, EXPERIENCE PLUS ACCESS, EXPERIENCE ACCESS, TRADE EXHIBITOR\n\nLanguage: ENGLISH\n\nFormat: IN-PERSON URL:https://sa2022.siggraph.org/en/full-program/?id=papers_157&sess=sess15 3 END:VEVENT END:VCALENDAR