BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Asia/Tokyo
X-LIC-LOCATION:Asia/Tokyo
BEGIN:STANDARD
TZOFFSETFROM:+0900
TZOFFSETTO:+0900
TZNAME:JST
DTSTART:18871231T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20250110T023312Z
LOCATION:Hall B7 (1)\, B Block\, Level 7
DTSTART;TZID=Asia/Tokyo:20241205T164100
DTEND;TZID=Asia/Tokyo:20241205T165300
UID:siggraphasia_SIGGRAPH Asia 2024_sess138_papers_629@linklings.com
SUMMARY:TextToon: Real-Time Text Toonify Head Avatar from Single Video
DESCRIPTION:Technical Papers\n\nLuchuan Song and Lele Chen (Univeristy of 
 Rochester), Celong Liu (Bytedance), Pinxin Liu (University of Rochester), 
 and Chenliang Xu (Univeristy of Rochester)\n\nWe propose TextToon, a metho
 d to generate a drivable toonified avatar. Given a short monocular video s
 equence and a written instruction about the avatar style, our model can ge
 nerate a high-fidelity toonified avatar that can be driven in real-time by
  another video with arbitrary identities. Existing related works heavily r
 ely on multi-view modeling to recover geometry via texture embeddings, pre
 sented in a static manner, leading to control limitations. The multi-view 
 video input also makes it difficult to deploy these models in real-world a
 pplications. To address these issues, we adopt a conditional embedding Tri
 -plane to learn realistic and stylized facial representations in a Gaussia
 n deformation field. Additionally, we expand the stylization capabilities 
 of 3D Gaussian Splatting by introducing an adaptive pixel-translation neur
 al network and leveraging patch-aware contrastive learning to achieve high
 -quality images. To push our work into consumer applications, we develop a
  real-time system that can operate at 48 FPS on a GPU machine and 15-18 FP
 S on a mobile machine. Extensive experiments demonstrate the efficacy of o
 ur approach in generating textual avatars over existing methods in terms o
 f quality and real-time animation. Please refer to our project page for mo
 re details: https://songluchuan.github.io/TextToon/.\n\nRegistration Categ
 ory: Full Access, Full Access Supporter\n\nLanguage Format: English Langua
 ge\n\nSession Chair: Hongbo Fu (Hong Kong University of Science and Techno
 logy)
URL:https://asia.siggraph.org/2024/program/?id=papers_629&sess=sess138
END:VEVENT
END:VCALENDAR