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:20250110T023313Z
LOCATION:Hall B7 (1)\, B Block\, Level 7
DTSTART;TZID=Asia/Tokyo:20241206T151900
DTEND;TZID=Asia/Tokyo:20241206T153100
UID:siggraphasia_SIGGRAPH Asia 2024_sess150_papers_169@linklings.com
SUMMARY:DiffH2O: Diffusion-Based Synthesis of Hand-Object Interactions fro
 m Textual Descriptions
DESCRIPTION:Technical Papers\n\nSammy Christen (ETH, Meta) and Shreyas Ham
 pali, Fadime Sener, Edoardo Remelli, Tomas Hodan, Eric Sauser, Shugao Ma, 
 and Bugra Tekin (Meta)\n\nWe introduce DiffH2O, a new diffusion-based fram
 ework for synthesizing realistic, dexterous hand-object interactions from 
 natural language. Our model employs a temporal two-stage diffusion process
 , dividing hand-object motion generation into grasping and interaction sta
 ges to enhance generalization to various object shapes and textual prompts
 . To improve generalization to unseen objects and increase output controll
 ability, we propose grasp guidance, which directs the diffusion model towa
 rds a target grasp, seamlessly connecting the grasping and interaction sta
 ges through a motion imputation mechanism.  We demonstrate the practical v
 alue of grasp guidance using hand poses extracted from images or grasp syn
 thesis methods. Additionally, we provide detailed textual descriptions for
  the GRAB dataset, enabling fine-grained text-based control of the model o
 utput. Our quantitative and qualitative evaluations show that DiffH2O gene
 rates realistic hand-object motions from natural language, generalizes to 
 unseen objects, and significantly outperforms existing methods on a standa
 rd benchmark and in perceptual studies.\n\nRegistration Category: Full Acc
 ess, Full Access Supporter\n\nLanguage Format: English Language\n\nSession
  Chair: Li-Yi Wei (Adobe Research)
URL:https://asia.siggraph.org/2024/program/?id=papers_169&sess=sess150
END:VEVENT
END:VCALENDAR
