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:20250110T023309Z
LOCATION:Hall B7 (1)\, B Block\, Level 7
DTSTART;TZID=Asia/Tokyo:20241203T150800
DTEND;TZID=Asia/Tokyo:20241203T151900
UID:siggraphasia_SIGGRAPH Asia 2024_sess108_papers_391@linklings.com
SUMMARY:ProcessPainter: Learning to draw from sequence data
DESCRIPTION:Technical Papers\n\nYiren Song (National University of Singapo
 re, Show Lab); Shijie Huang, Chen Yao, and Hai Ci (National University of 
 Singapore); Xiaojun Ye (Zhejiang University); Jiaming Liu (Tiamat); Yuxuan
  Zhang (Shanghai Jiao Tong University); and Mike Zheng Shou (National Univ
 ersity of Singapore)\n\nThe painting process of artists is inherently step
 wise and varies significantly among different painters and styles. Generat
 ing detailed, step-by-step painting processes is essential for art educati
 on and research, yet remains largely underexplored. Traditional stroke-bas
 ed rendering methods break down images into sequences of brushstrokes, yet
  they fall short of replicating the authentic processes of artists, with l
 imitations confined to basic brushstroke modifications. Text-to-image mode
 ls utilizing diffusion processes generate images through iterative denoisi
 ng, also diverge substantially from artists' painting process. To address 
 these challenges, we introduce ProcessPainter, a text-to-video model that 
 is initially pre-trained on synthetic data and subsequently fine-tuned wit
 h a select set of artists' painting sequences using the LoRA model. This a
 pproach successfully generates painting processes from text prompts for th
 e first time. Furthermore, we introduce an Artwork Replication Network cap
 able of accepting arbitrary-frame input, which facilitates the controlled 
 generation of painting processes, decomposing images into painting sequenc
 es, and completing semi-finished artworks. This paper offers new perspecti
 ves and tools for advancing art education and image generation technology.
  Our code is available at: \url{https://github.com/nicolaus-huang/ProcessP
 ainter}\n\nRegistration Category: Full Access, Full Access Supporter\n\nLa
 nguage Format: English Language\n\nSession Chair: I-Chao Shen (The Univers
 ity of Tokyo)
URL:https://asia.siggraph.org/2024/program/?id=papers_391&sess=sess108
END:VEVENT
END:VCALENDAR