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:20260817T171530Z
LOCATION:Hall B7 (1)\, B Block\, Level 7
DTSTART;TZID=Asia/Tokyo:20241203T144500
DTEND;TZID=Asia/Tokyo:20241203T145600
UID:siggraphasia_SIGGRAPH Asia 2024_sess108_papers_1187@linklings.com
SUMMARY:HFH-Font: Few-shot Chinese Font Synthesis with Higher Quality, Fas
 ter Speed, and Higher Resolution
DESCRIPTION:Hua Li (Wangxuan Institute of Computer Technology, Peking Univ
 ersity) and Zhouhui Lian (Wangxuan Institute of Computer Technology, Pekin
 g University; State Key Laboratory of General Artificial Intelligence, Pek
 ing University)\n\nThe challenge of automatically synthesizing high-qualit
 y vector fonts, particularly for writing systems (e.g., Chinese) consistin
 g of huge amounts of complex glyphs, remains unsolved. Existing font synth
 esis techniques fall into two categories: 1) methods that directly generat
 e vector glyphs, and 2) methods that initially synthesize glyph images and
  then vectorize them. However, the first category often fails to construct
  complete and correct shapes for complex glyphs, while the latter struggle
 s to efficiently synthesize high-resolution (i.e., 1024 × 1024 or higher) 
 glyph images while preserving local details. In this paper, we introduce H
 FH-Font, a few-shot font synthesis method capable of efficiently generatin
 g high-resolution glyph images that can be converted into high-quality vec
 tor glyphs. More specifically, our method employs a diffusion model-based 
 generative framework with component-aware conditioning to learn different 
 levels of style information adaptable to varying input reference sizes. We
  also design a distillation module based on Score Distillation Sampling fo
 r 1-step fast inference, and a style-guided super-resolution module to ref
 ine and upscale low-resolution synthesis results. Extensive experiments, i
 ncluding a user study with professional font designers, have been conducte
 d to demonstrate that our method significantly outperforms existing font s
 ynthesis approaches. Experimental results show that our method produces hi
 gh-fidelity, high-resolution raster images which can be vectorized into hi
 gh-quality vector fonts. Using our method, for the first time, large-scale
  Chinese vector fonts of a quality comparable to those manually created by
  professional font designers can be automatically generated.\n\nRegistrati
 on Category: Full Access, Full Access Supporter\n\nLanguage Format: Englis
 h Language\n\nSession Chair: I-Chao Shen (The University of Tokyo)\n\n
URL:https://asia.siggraph.org/2024/program/?id=papers_1187&sess=sess108
END:VEVENT
END:VCALENDAR