A ray tracing solution for diffuse interreflection

Abstract

Cre gory J. Ward Franc" M. Rdin , ft,1 'n Rob ert D. Clt arLigAting S1, tem, R u tGrcA Lawr enee Ber keley L aboratory 1 e ye/atrllft Rd ., 9().3111 B erl:t.lt,1 , CA 94 7£0 ( / 15) , S6 -, 757 Abs tn .ct An efficient. ray tr acing met hod is pre se nted for calcul ating inte rrefleerions betw een s ur faces wit h bot h diffuse an d spec ulai' co mpone nts. A Mont e Car lo t echn ique co mp utes t he indi rect co nt r ibutions to illum inan ce a.t locat ions chosen by th e rendering proc ess. Th e ind irect illu m inan ce values ar e a verage d ove r surfac es and used in place of a cOM ta n t " amb ient." te rm . Illumin an ce calculations are mad e only (or th ose are as par t icipati ng in t he selec ted vi ew, an d the results are sto red so t hat subse quent views can re use co mmon valu es . Th e den sity of th e cal culati on is ad ju ste d to m aintain a cons tan t. ac cu racy , perm ittin g less popu lated po rt icna of t he sce ne to be co mpute d quickly . Succ essive reflections use pro por tional ly fewer sa mples , whi ch spe eds t he process and pro vides a natur al lim it to recurs ion . Th e tec hni que can also mode l diffuse t rans mission and illumina t ion fro m lar ge aru $()U fCe8, such as t he sky .Ge neral T erms : Algorithm , com plexity .Add it ion al K eyword s a nd Ph rases : Cac hing, diffuse, illu minan ce, int erre ftectio n , lu minan ce, Mo nte Car lo tec hnique , rad iosity , ray t rac ing, rend erin g, spec ular . 1. In tr oduc.tion © 1988 ACM .(}.. 89191·275-6/ 88/008/0085 $00. 75 85 R o =. "average"distanceto surfaces at PoTh e ch~ge in % becom es th e di stance betw een two pointe, and t he chan ge In ebecomes th e angl e betw een two eurfeee norm als. Th is equation is use d to est imate t he rela t ive change in illuminance for any geometry . Bo t h th e po ints and t he surface nor mals~determin ed by t.he ray interse ct-i on calcu latio n . R 0 is the harm on ic mean (reciprocal mean reciproca l) of distance!l to visible surfaces, which can be computed (rom ray lengt hs during prim ary eval uat ion . 87 10 . NI8bit.a, TOmoyuD and Eihacbiro Nakamae , uCoatlo\lOUl TOIle Repneentatioo. 01 Tbree--Dimenaioaal Objectc Taki.a.1 A«ount 01 SbadOWfland Iaterre 8ectioD," 0 0".

Selective rendering

“…Although Radiance uses distributed ray tracing to render images, the irradiance cache [Ward et al 1988] is used to accelerate the calculation of the indirect diffuse component. As the irradiance cache is a shared data structure, it is non-trivial to parallelise.…”

“…The view independent nature of the indirect diffuse component of global illumination was exploited by Ward et al to sufficiently populate the innovative irradiance cache, from which other samples could interpolate with significant computational cost savings [Ward et al 1988]. Other component based rendering have included [Sillion and Puech 1989;Shirley 1990;Heckbert 1990;Chen et al 1991;Stokes et al 2004].…”

Radiance cache splatting

“…Our work is based on the irradiance and radiance (or (ir)radiance) caching algorithms [Ward et al 1988;Křivánek et al 2005b]. These approaches rely on sparse sampling and interpolation of indirect incoming radiance.…”

“…We propose an image-plane method which uses the opposite approach: for a given record, our algorithm determines which visible points it contributes to. This set of visible points is determined using the weighting function and interpolation scheme proposed in [Ward et al 1988]: a given record k located at p k can contribute to the incoming radiance at point p only if p − p k ≤ aR k . a represents a user-defined accuracy value and R k is the harmonic mean distance to the objects visible from p k .…”

Improved radiance gradient computation

“…Irradiance caching [Ward et al 1988;Křivánek et al 2005] is a ray tracing-based method for computing indirect diffuse illumination. It exploits the indirect illumination coherence by sparsely sampling, caching and interpolating the indirect diffuse illumination over surfaces, instead of computing it independently for each pixel.…”