160 likes | 291 Views
Distributed Ray Tracing Part 2. 黃聰賢. Overview. Render Equation BRDF Importance Sampling Implementation. Rendering Equation (1). ω o. x. is the radiance from a point to given direction w o. Rendering Equation (2). ω o. x. is the emitted radiance.
E N D
Overview • Render Equation • BRDF • Importance Sampling • Implementation
Rendering Equation (1) ωo x • is the radiance from a point to given direction wo
Rendering Equation (2) ωo x • is the emitted radiance • is non-zero if x is emissive(a light source)
Rendering Equation (3) ωi ωo x • Sum of the contributionfrom all of the other direction in the scene
Rendering Equation (4) ωi ωo x • Radiance from all hemisphere direction
Integration over hemisphere y0 ω0 y1 normal ω1 eye yi ωi x Spherical sample direction L(x,wo) = (2 PI / #samples) * ∑ [BRDF(x,wo,wi)*L(yi,-wi) * cos(n,ωi)]
Spherical Uniform Sampling Generate two uniform random variables in [0,1) : ξx, ξy x = sin(θ) cos(φ) y = sin(θ) sin(φ) z = cos(θ) φ
Why? Too Many Too Coarse Importance
Implement of Importance Sampling • Generate enough samples (uniform samples) • Compute the importance of each sample • Build the CDF of importance • Generate uniform random variables over [0,1) • Use Inverse CDF to choose a sample • Divide the contribution of each sample by its probability
Direct Lighting • Use Phong Lighting Model. • Add the lighting effect if visibility is one. I * (Kd * dot(N, L) + Ks * pow(dot(E, R), Ns) ) N E L R
Indirect Lighting • Use importance sampling to choose direction • If the direction hits a point yi ,compute the yi direct lighting y0 ω0 y1 normal ω1 eye yi ωi x
L(x, ωo) = (2 PI / #samples) * ∑ [BRDF(x, ωo, ωi)*L(yi,-ωi) * cos(n,ωi)] L(x, ωo) = (1.0 / #samples) * ∑ { L(yi ,-ωi) * [Kd * dot(ωi, N) + Ks * pow(dot(E, reflect(ωi, N)), Ns) ] } N E yi ωi x