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_558@linklings.com SUMMARY:Production-Ready Face Re-Aging for Visual Effects DESCRIPTION:Technical Papers\n\nProduction-Ready Face Re-Aging for Visual Effects\n\nZoss, Chandran, Sifakis, Gross, Gotardo...\n\nPhotorealistic di gital re-aging of faces in video is becoming increasingly common in entert ainment and advertising. But the predominant 2D painting workflow often r equires frame-by-frame manual work that can take days to accomplish, even by skilled artists. Although research on facial image re-aging has attemp ted to automate and solve this problem, current techniques are of little p ractical use as they typically suffer from facial identity loss, poor reso lution, and unstable results across subsequent video frames. In this paper , we present the first practical, fully-automatic and production-ready met hod for re-aging faces in video images. Our first key insight is in addre ssing the problem of collecting longitudinal training data for learning to re-age faces over extended periods of time, a task that is nearly impossi ble to accomplish for a large number of real people. We show how such a lo ngitudinal dataset can be constructed by leveraging the current state-of-t he-art in facial re-aging that, although failing on real images, do provid e photoreal re-aging results on synthetic faces. Our second key insight is then to leverage such synthetic data and formulate facial re-aging as a p ractical image-to-image translation task that can be performed by training a well-understood U-Net architecture, without the need for more complex n etwork designs. We demonstrate how the simple U-Net leads to surprisingly better results for re-aging real faces on video, with unprecedented tempor al stability and preservation of facial identity across variable viewpoint s and lighting conditions. Finally, our new face re-aging network (FRAN) i ncorporates simple and intuitive mechanisms that provides artists with loc alized control and creative freedom to direct and fine-tune the re-aging e ffect, a feature that is largely important in real production pipelines an d often overlooked in related research work.\n\nRegistration Category: FUL L ACCESS, EXPERIENCE PLUS ACCESS, EXPERIENCE ACCESS, TRADE EXHIBITOR\n\nLa nguage: ENGLISH\n\nFormat: IN-PERSON URL:https://sa2022.siggraph.org/en/full-program/?id=papers_558&sess=sess15 3 END:VEVENT END:VCALENDAR