A K no wle dg e Ba s e d Appro a ch to Mo de ling Po r tr a it Pa inting Methodology Ste ve DiPao la Schoo l of I nte ract ive Art a nd Te chn ology Simon Fraser University 250- 13450 102 nd Ave. Surrey, British Columbia Canada [email protected] http://www.dipaola.org Tr a d itio na l p ortra it a rtis ts use a s p ec ific b u t o p en h uman v is io n me th o d o lo g y to crea t e a painterly portrait of a live or photographed sitter. Portrait artists attempt to simplify, compose and leave out what’s irrelevant, emphasizing what’s important. While seemingly a qualitative pursuit, artists use known but open techniques such as relying on source tone ov e r co lour to indir e ct into a co lour te m pe rature mo de l, using " sh arpne ss " to create a centre of interest, using edges to move the viewers gaze, and other techniques to filter and emphasize. Our interdisciplinary work attempts to compile and incorporate this portrait painte r knowle dge into a multi -space param e terize d sys tem that can cre ate an array of painterly rendering output. INTRODUCTION Non Photorealistic Rendering (NPR) is a computer graphics technique which creates imagery with a wide variety of expressive styles inspired by painting, drawing, technical illustration, and cartoons. This is in contrast to typical computer graphics, which focuses on photorealism. NPR already has applications in video games, movies, architectural and technical illustration, animation and rising fields such as computational photography. NPR also has applications in learning and medicine (e.g. communication systems for autistic children) where filtering out unnecessary detail is important. Man y current NPR te chni ques rely on computer im a ging a pproaches (e.g. edg e detection, im ag e se gm en tati on) t ha t m odel at the physical leve l such as blobs, str okes and lines. We propose a novel approach to painterly rendering which relies on pa ram e te ri zing a cog nitive knowled ge s pa ce of how a hu m an p a inter pa ints – i.e. their ope n m et hodo logy to the proces s. For i nsta nce , unlike m ost NPR te chni que s a n artist holds a semantic representation of the object to be rendered (e.g. a face for a portrait, or an em otiona l em pha sis). I n ge ne ral, artistic m et hodo logy a tte m pts th e following: from the phot ogra ph or live si tte r, the pa inti ng m ust si m plify, com pose a nd leave out wha t’s irrelevan t, em pha sizi ng wha t’s im porta nt Since h um an pa inte rs ha ve knowledge of the source i m ag e ry, we a re lim iting this a pproa ch to portr a iture a nd th e ref ore ta ke advantage of portrait and facial knowledge in the NPR process, as well as portraiture's strong foreground versus background distinct.. We believe this portrait painting knowledge approach when fully realized has two intertwining and interdisciplinary benefits. The first benefit is creating a novel type of NPR system that within its domain (portraiture but with benefits to still life and figurative work) may produce both a wider range of results an d im proved results com pa red to curren t techn ique s. The second possible benefit, which is more speculative, is that portrait artists over thousands of yea rs ha ve som ewh a t uncon scious ly evolved a ‘pa inti ng m e thod ology’ e xploiting specific human vision and cognitive functions, which, if presented in a quantitative way,