Effcient Acquisition and Realistic Rendering of Car Paint

Johannes Günther, Tongbo Chen, Michael Goesele, Ingo Wald, and Hans-Peter Seidel

Abstract

The outside appearance of cars is mostly defined through only two distinct materials – glass and car paint. While glass can rather easily be simulated by the simple physical laws of reflection and refraction, modelling car paint is more challenging.

In this paper we present a framework for the efficient acquisition and realistic rendering of real-world car paint. This is achieved by building an easy-to-reproduce measuring setup, fitting the measured data to a general BRDF model for car paint, adding a component for simulating the sparkling effect of metallic paints, and rendering using a specially designed shader in a realtime ray tracer.

Paper

[pdf][bib]

VMV 2005
9 pages
8233 kb

bibtex entry

[ppt]

Slides
9318 kb

BRDF Data

Here we provide our measured BRDF data sets.

Licence

You may download and use the data only for non-commercial and research purposes!

Format description

The 3-dimensional BRDF is stored using the halfway vector based parameterization (see Section 4.1 in the paper for details). φ_d is split into 180 bins, while θ_h and θ_d are subdivided into 90 bins. The order of the dimensions is φ_d, θ_h, θ_d. Each table entry consists of 3 float values, representing the red, green and blue channel.
This is more or less the raw measured data. The BRDF values are not normalized (there can be values grater than 1.0) and no color calibration has been done. Invalid values are marked with -1.0.

Example code

#include <fstream>
using namespace std;

typedef struct {
  float r, g, b;
} Color;

const unsigned THETA = 90;
const unsigned PHI = 180;
Color *data = new Color[THETA * THETA * PHI];
    
ifstream file("brdf_data");
// skip header
file.seekg(84);

Color col;
for (unsigned phi_d = 0; phi_d < PHI; phi_d++)
  for (unsigned theta_h = 0; theta_h < THETA; theta_h++)
    for (unsigned theta_d = 0; theta_d < THETA; theta_d++) {
      fread(&col, sizeof(Color), 1, file);
      // now a different order of the dimensions can be used
      data[(theta_h * THETA + theta_d) * PHI + phi_d] = col;
    }

Data sets

white [gz][zip]	2k_acrylack [gz][zip]	blue [gz][zip]	blue_matte [gz][zip]
ford_f8 [gz][zip]	BMW339 [gz][zip]	OPEL_titan [gz][zip]	polaris_silver [gz][zip]
all sets [tar.gz][zip]	8 data sets 58 Mb