BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Australia/Melbourne
X-LIC-LOCATION:Australia/Melbourne
BEGIN:DAYLIGHT
TZOFFSETFROM:+1000
TZOFFSETTO:+1100
TZNAME:AEDT
DTSTART:19721003T020000
RRULE:FREQ=YEARLY;BYMONTH=4;BYDAY=1SU
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:19721003T020000
TZOFFSETFROM:+1100
TZOFFSETTO:+1000
TZNAME:AEST
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20240214T070247Z
LOCATION:Meeting Room C4.9+C4.10\, Level 4 (Convention Centre)
DTSTART;TZID=Australia/Melbourne:20231214T110000
DTEND;TZID=Australia/Melbourne:20231214T111000
UID:siggraphasia_SIGGRAPH Asia 2023_sess152_papers_639@linklings.com
SUMMARY:Anything to Glyph: Artistic Font Synthesis via Text-to-Image Diffu
 sion Model
DESCRIPTION:Technical Papers\n\nChangShuo Wang, Lei Wu, XiaoLe Liu, and Xi
 ang Li (Shandong University); Lei Meng (Shandong University, Shandong Rese
 arch Institute of Industrial Technology); and Xiangxu Meng (Shandong Unive
 rsity)\n\nThe automatic generation of artistic fonts is a challenging task
  that attracts many research interests. Previous methods specifically focu
 s on glyph or texture style transfer. However, we often come across creati
 ve fonts composed of objects in posters or logos. These fonts have proven 
 to be a challenge for existing methods as they struggle to generate simila
 r designs. This paper proposes a novel method for generating creative arti
 stic fonts using a pre-trained text-to-image diffusion model. Our model ta
 kes a shape image and a prompt describing an object as input and generates
  an artistic glyph image consisting of such objects. Specifically, we intr
 oduce a novel heatmap-based weak position constraint method to guide the p
 ositioning of objects in the generated image, and we also propose the Late
 nt Space Semantic Augmentation Module that improves other information whil
 e constraining object position. Our approach is unique in that it can pres
 erve the object's original shape while constraining its position. And our 
 training method requires only a small quantity of generated data, making i
 t an efficient unsupervised learning approach. Experimental results demons
 trate that our method can generate various glyphs, including Chinese, Engl
 ish, Japanese, and symbols, using different objects. We also conducted qua
 litative and quantitative comparisons with various position control method
 s for the diffusion model. The results indicate that our approach outperfo
 rms other methods in terms of visual quality, innovation, and user evaluat
 ion.\n\nRegistration Category: Full Access\n\nSession Chair: Haisen Zhao (
 Shandong University)
URL:https://asia.siggraph.org/2023/full-program?id=papers_639&sess=sess152
END:VEVENT
END:VCALENDAR