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 B5 (2)\, B Block\, Level 5 DTSTART;TZID=Asia/Tokyo:20241205T163000 DTEND;TZID=Asia/Tokyo:20241205T174000 UID:siggraphasia_SIGGRAPH Asia 2024_sess137@linklings.com SUMMARY:Diffuse and Conquer DESCRIPTION:Technical Papers\n\nEach Paper gives a 10 minute presentation. \n\nPortrait Video Editing Empowered by Multimodal Generative Priors\n\nWe introduce PortraitGen, a powerful portrait video editing method that achi eves consistent and expressive stylization with multimodal prompts. Tradit ional portrait video editing methods often struggle with 3D and temporal c onsistency, and typically lack in rendering quality and efficiency. To add re...\n\n\nXuan Gao, Haiyao Xiao, Chenglai Zhong, Shimin Hu, Yudong Guo, a nd Juyong Zhang (University of Science and Technology of China)\n--------- ------------\nMV2MV: Multi-View Image Translation via View-Consistent Diff usion Models\n\nImage translation has various applications in computer gra phics and computer vision, aiming to transfer images from one domain to an other. Thanks to the excellent generation capability of diffusion models, recent single-view image translation methods achieve realistic results. Ho wever, directly appl...\n\n\nYoucheng Cai, Runshi Li, and Ligang Liu (Univ ersity of Science and Technology of China)\n---------------------\nStyleCr after: Taming Stylized Video Diffusion with Reference-Augmented Adapter Le arning\n\nText-to-video (T2V) models have shown remarkable capabilities in generating diverse videos. However, they struggle to produce user-desired artistic videos due to (i) text's inherent clumsiness in expressing speci fic styles and (ii) the generally degraded style fidelity. To address thes e challenges, ...\n\n\nGongye Liu (Tsinghua University); Menghan Xia, Yong Zhang, and Haoxin Chen (Tencent AI lab); Jinbo Xing (Chinese University o f Hong Kong); Yibo Wang (Tsinghua University); Xintao Wang and Ying Shan ( Tencent); and Yujiu Yang (Tsinghua University)\n---------------------\nSta bleNormal: Reducing Diffusion Variance for Stable and Sharp Normal\n\nThis work addresses the challenge of high-quality surface normal estimation fr om monocular colored inputs (i.e., images and videos), a field which has r ecently been revolutionized by repurposing diffusion priors. However, prev ious attempts still struggle with stochastic inference, conflicting with t ...\n\n\nChongjie Ye and Lingteng Qiu (FNii, The Chinese University of Hon g Kong, Shenzhen; SSE, The Chinese University of Hong Kong, Shenzhen); Xia odong Gu and Qi Zuo (Alibaba); Yushuang Wu (FNii, The Chinese University o f Hong Kong, Shenzhen; SSE, The Chinese University of Hong Kong, Shenzhen) ; Zilong Dong and Liefeng Bo (Alibaba); Yuliang Xiu (Max Planck Institute for Intelligent Systems); and Xiaoguang Han (SSE, The Chinese University o f Hong Kong, Shenzhen; FNii, The Chinese University of Hong Kong, Shenzhen )\n---------------------\nHyperGAN-CLIP: A Unified Framework for Domain Ad aptation, Image Synthesis and Manipulation\n\nGenerative Adversarial Netwo rks (GANs), particularly StyleGAN and its variants, have demonstrated rema rkable capabilities in generating highly realistic images. Despite their s uccess, adapting these models to diverse tasks such as domain adaptation, reference-guided synthesis, and text-guided manipu...\n\n\nAbdul Basit Ane es (Koç University), Ahmet Canberk Baykal (University of Cambridge), Muham med Burak Kizil (Koç University), Duygu Ceylan (Adobe Research), Erkut Erd em (Hacettepe University), and Aykut Erdem (Koç University)\n------------- --------\nFast High-Resolution Image Synthesis with Latent Adversarial Dif fusion Distillation\n\nDiffusion models are the main driver of progress in image and video synthesis, but suffer from slow inference speed. Distilla tion methods, like the recently introduced adversarial diffusion distillat ion (ADD) aim to shift the model from many-shot to single-step inference, albeit at the cost of expen...\n\n\nAxel Sauer, Frederic Boesel, Tim Dockh orn, Andreas Blattmann, Patrick Esser, and Robin Rombach (Black Forest Lab s)\n\nRegistration Category: Full Access, Full Access Supporter\n\nLanguag e Format: English Language\n\nSession Chair: Michael Rubinstein (Google) END:VEVENT END:VCALENDAR