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:20231214T105000
DTEND;TZID=Australia/Melbourne:20231214T110000
UID:siggraphasia_SIGGRAPH Asia 2023_sess152_papers_810@linklings.com
SUMMARY:Text-Guided Vector Graphics Customization
DESCRIPTION:Technical Papers\n\nPeiying Zhang (City University of Hong Kon
 g), Nanxuan Zhao (Adobe Research), and Jing Liao (City University of Hong 
 Kong)\n\nVector graphics are widely used in digital art and valued by desi
 gners for their scalability and layer-wise topological properties. However
 , the creation and editing of vector graphics necessitate creativity and d
 esign expertise, leading to a time-consuming process. In this paper, we pr
 opose a novel pipeline that generates high-quality customized vector graph
 ics based on\ntextual prompts while preserving the properties and layer-wi
 se information of a given exemplar SVG. Our method harnesses the capabilit
 ies of large pre-trained text-to-image models. By fine-tuning the cross-at
 tention layers of the model, we generate customized raster images guided b
 y textual prompts. To initialize the SVG, we introduce a semantic-based pa
 th alignment method that preserves and transforms crucial paths from the e
 xemplar SVG. Additionally, we optimize path parameters using both image-le
 vel and vector-level losses, ensuring smooth shape deformation while align
 ing with the customized raster image. We extensively evaluate our method u
 sing multiple metrics from vector-level, image-level, and text-level persp
 ectives. The evaluation results demonstrate the effectiveness of our pipel
 ine in generating diverse customizations of vector graphics with exception
 al quality. The project page is https://intchous.github.io/SVGCustomizatio
 n.\n\nRegistration Category: Full Access\n\nSession Chair: Haisen Zhao (Sh
 andong University)
URL:https://asia.siggraph.org/2023/full-program?id=papers_810&sess=sess152
END:VEVENT
END:VCALENDAR