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:20241205T131400
DTEND;TZID=Asia/Tokyo:20241205T132800
UID:siggraphasia_SIGGRAPH Asia 2024_sess132_papers_249@linklings.com
SUMMARY:Monkey See, Monkey Do: Harnessing Self-attention in Motion Diffusi
 on for Zero-shot Motion Transfer
DESCRIPTION:Technical Papers\n\nSigal Raab, Inbar Gat, Nathan Sala, Guy Te
 vet, and Rotem Shalev-Arkushin (Tel Aviv University); Ohad Fried (Reichman
  University); and Amit Haim Bermano and Daniel Cohen-Or (Tel Aviv Universi
 ty)\n\nGiven the remarkable results of motion synthesis with diffusion mod
 els, a natural question arises: how can we effectively leverage these mode
 ls for motion editing? Existing diffusion-based motion editing methods ove
 rlook the profound potential of the prior embedded within the weights of p
 re-trained models, which enables manipulating the latent feature space; he
 nce, they primarily center on handling the motion space. In this work, we 
 explore the attention mechanism of pre-trained motion diffusion models. We
  uncover the roles and interactions of attention elements in capturing and
  representing intricate human motion patterns, and carefully integrate the
 se elements to transfer a leader motion to a follower one while maintainin
 g the nuanced characteristics of the follower, resulting in zero-shot moti
 on transfer. Manipulating features associated with selected motions allows
  us to confront a challenge observed in prior motion diffusion approaches,
  which use general directives (e.g., text, music) for editing, ultimately 
 failing to convey subtle nuances effectively. Our work is inspired by how 
 a monkey closely imitates what it sees while maintaining its unique motion
  patterns; hence we call it Monkey See, Monkey Do, and dub it MoMo. Employ
 ing our technique enables accomplishing tasks such as synthesizing out-of-
 distribution motions, style transfer, and spatial editing. Furthermore, di
 ffusion inversion is seldom employed for motions; as a result, editing eff
 orts focus on generated motions, limiting the editability of real ones. Mo
 Mo harnesses motion inversion, extending its application to both real and 
 generated motions. Experimental results show the advantage of our approach
  over the current art. In particular, unlike methods tailored for specific
  applications through training, our approach is applied at inference time,
  requiring no training. Our webpage, https://monkeyseedocg.github.io, incl
 udes links to videos and code.\n\nRegistration Category: Full Access, Full
  Access Supporter\n\nLanguage Format: English Language\n\nSession Chair: Y
 i Zhou (Adobe)
URL:https://asia.siggraph.org/2024/program/?id=papers_249&sess=sess132
END:VEVENT
END:VCALENDAR