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:20260817T171534Z
LOCATION:Hall B7 (1)\, B Block\, Level 7
DTSTART;TZID=Asia/Tokyo:20241203T130000
DTEND;TZID=Asia/Tokyo:20241203T141000
UID:siggraphasia_SIGGRAPH Asia 2024_sess105@linklings.com
SUMMARY:Make It Yours - Customizing Image Generation
DESCRIPTION:Each Paper gives a 10 minute presentation.\n\nMoA: Mixture-of-
 Attention for Subject-Context Disentanglement in Personalized Image Genera
 tion\n\nWe introduce a new architecture for personalization of text-to-ima
 ge diffusion models, coined Mixture-of-Attention (MoA). Inspired by the Mi
 xture-of-Experts mechanism utilized in large language models (LLMs), MoA d
 istributes the generation workload between two attention pathways: a perso
 nalized bran...\n\n\nKuan-Chieh Wang, Daniil Ostashev, Yuwei Fang, Sergey 
 Tulyakov, and Kfir Aberman (Snap Inc.)\n---------------------\nReVersion: 
 Diffusion-Based Relation Inversion from Images\n\nDiffusion models gain in
 creasing popularity for their generative capabilities. Recently, there hav
 e been surging needs to generate customized images by inverting diffusion 
 models from exemplar images, and existing inversion methods mainly focus o
 n capturing object appearances (i.e., the "look"). How...\n\n\nZiqi Huang,
  Tianxing Wu, Yuming Jiang, Kelvin C.K. Chan, and Ziwei Liu (S-Lab for Adv
 anced Intelligence, Nanyang Technological University Singapore)\n---------
 ------------\nPALP: Prompt Aligned Personalization of Text-to-Image Models
 \n\nContent creators often aim to create personalized images using persona
 l subjects that go beyond the capabilities of conventional text-to-image m
 odels. Additionally, they may want the resulting image to encompass a spec
 ific location, style, ambiance, and more. Existing personalization methods
  may com...\n\n\nMoab Arar (Tel Aviv University), Andrey Voynov and Amir H
 ertz (Google Research), Omri Avrahami (Hebrew University of Jerusalem), Sh
 lomi Fruchter and Yael Pritch (Google Research), Daniel Cohen-Or (Tel Aviv
  University), and Ariel Shamir (Reichman University)\n--------------------
 -\nCustomizing Text-to-Image Models with a Single Image Pair\n\nArt reinte
 rpretation is the practice of creating a variation of a reference work, ma
 king a paired artwork that exhibits a distinct artistic style. We ask if s
 uch an image pair can be used to customize a generative model to capture t
 he demonstrated stylistic difference. We propose Pair Customization,...\n\
 n\nMaxwell Jones, Sheng-Yu Wang, and Nupur Kumari (Carnegie Mellon Univers
 ity); David Bau (Northeastern University); and Jun-Yan Zhu (Carnegie Mello
 n University)\n---------------------\nCustomizing Text-to-Image Diffusion 
 with Object Viewpoint Control\n\nModel customization introduces new concep
 ts to existing text-to-image models, enabling the generation of these new 
 concepts/objects in novel contexts.\nHowever, such methods lack accurate c
 amera view control with respect to the new object, and users must resort t
 o prompt engineering (e.g., adding "to...\n\n\nNupur Kumari and Grace Su (
 Carnegie Mellon Uniersity); Richard Zhang, Taesung Park, and Eli Shechtman
  (Adobe Research); and Jun-Yan Zhu (Carnegie Mellon Uniersity)\n----------
 -----------\nIdentity-Preserving Face Swapping via Dual Surrogate Generati
 ve Models\n\nIn this study, we revisit the fundamental setting of face-swa
 pping models and reveal that only using implicit supervision for training 
 leads to the difficulty of advanced methods to preserve the source identit
 y. We propose a novel reverse pseudo-input generation approach to offer su
 pplemental data f...\n\n\nZiyao Huang and Fan Tang (Institute of Computing
  Technology, Chinese Academy of Sciences); Yong Zhang (Tencent); Juan Cao,
  Chengyu Li, Sheng Tang, and Jintao Li (Institute of Computing Technology,
  Chinese Academy of Sciences); and Tong-Yee Lee (National Cheng Kung Unive
 rsity)\n\nRegistration Category: Full Access, Full Access Supporter\n\nLan
 guage Format: English Language\n\nSession Chair: Kfir Aberman (Decart AI)
END:VEVENT
END:VCALENDAR