This is the second version of a simple ray tracer written here at
RPI. The second version expanded its capabilities with the inclusion of
distributed ray tracing. Detailed information on what distributed ray
tracing is can be found in the paper by Cook, Porter and Carpenter:
"Distributed Ray Tracing", ACM Computer Graphics, Vol.18, Num.3, July 1984.
New capabilities include: Gloss (blurred reflection), translucency
(blurred refraction), penumbras (area light sources), motion blur, square
intersection routine and a field of view options. The eye can be in any
position now and can look towards any direction. Antialising can also be
done using stochastic sampling. The old version was also supporting
reflections, refractions and shadows using spheres. Here is a list
of the files, and what does each file contain:
bg.c: bgcolor() evaluates the background color for a given
ray.
initialize.c: initialize() does some useful setup.
intersect.c: sphere() Intersection routine with a sphere.
quad() Intersection routine with a square.
intersect() Main intersection routine (calls sphere()
).
main.c: main() Main body of the program.
readfile.c: readfile() Reads in the input data.
shade.c: shadow() Calculates the existance of a shadow ray.
reflect() Find the reflection vector.
refract() Find the refraction vector.
shade() Calculate Phong's shading function.
trace.c: quickcos() Calculate a fast cos between 0 and pi / 2.
quickinvcos() Calculate a fast inverse cosine.
rnd() Random number generator between 0 and 1.
rand1() Random number generator between -1 and 1.
grand() Approximate gaussian random number
generator. sample_ray() Take a ray somewhere inside a
solid angle. trace_a_ray() Trace a single ray.
trace() Trace a single ray inside a solid angle.
raytrace() Ray trace the whole picture.
vector.c: vadd() vector addition.
vsub() vector subtraction.
vneg() vector negation.
svproduct() scalar - vector product.
vdot() dot product.
vcross() cross product.
norm() normalize a vector.
ray.h: Include file for every file used in the raytracer.
vector.h: Include file for every file using vectors.
The ray tracer is written so it can be easily understood (at least
that version), and it is fully commented. Nevertheless, probably it won't
be understood by a newcomer.
The format of the input file is as follows:
<fov> <eye> <dir> <up> <time> <background> <iter> <light> <nos> <nosq> x y
z r [ambient] [diff] [spec] refl r g b refr r g b width index
refl_diffuse refr_diffuse tx ty tz x y z x y z x y z [ambient]
[diff] [spec] refl r g b refr r g b width index refl_diffuse
refr_diffuse tx ty tz
where:
fov field of view eye x y z components of the eye
position dir x y z components of the eye direction
up x y z components of the up vector time start and
end time of the picture background a character specifying the
background cuing as follows: n: no cuing. Background has a
constant intensity of 0.2 x: the intensity of the
background depends of the x direction. y: the intensity of
the background depends of the y direction. z: the
intensity of the background depends of the z direction.
iter number of samples per pixel. light x y z
components and solid angle of the light source. nos number of
spheres nosq number of squares [ambient] r g b components
of ambient [diff] r g b components of diffuse [spec] r g
b components of specular refl r g b reflection ratio and color of the
reflection refr r g b refraction ratio and color of the refraction
width specular width exponent index index of refraction
refl_diffuse angle of diffusion when reflecting refr_diffuse angle of
diffusion when refracting tx ty tz velocities on the specified axes.
Hints: Each frame is a snapsot of a given time length. The time
limits are specified in the input file. Each object has the capability to
move during that time in a strait line. Motion blur is then observed. If
you specify only one sample per pixel, then the blur won't be so good since
it evaluates the color of the ray with only one try. The more samples the
better, altough anything more than 20 or 30 doesn't do any good. 5 is a
good approximation. You can produce penumbras by specifying an angle in
the light source. That deviates from real life where each life has an area
and the further away you are from the light source, the smaller the
penumbras. Here, the size of the blurred shadow does not depend on the
distance from the light source. The refl_diffuse and refr_diffuse produce
non-sharp reflections and refractions. The argument is in degrees.
Anything less than 15 or 20 is good, altough the closer to zero the better
you can see it.
Known bugs: Polygons appear completely shadowd if the order of the
vertices is not right. I always forget which is the right order.
---- The square is defined by only 3 of its points. The first and
the second vertices specify one of the sides, and the first and third
specify the second. If you assume that second and third specify a side,
then you probably won't get the right result.
The format of the output file is simple. In the beginning there
are 2 integers (that can be read with fread() on a SUN) showing xsize and
ysize of the picture. After that follow the pixels in scan-line order.
Each pixel uses 3 bytes (one for red, green and blue), totalling 16777216
colors. You can change the format of that file to tailor your needs. It
can be done easily by changing the funcion raytrace() in file trace.c
If you want you can inform me with any bugs that the program might
have or any features that you want the upcoming versions to have.
Good luck!
George Kyriazis kyriazis@turing.cs.rpi.edu
kyriazis@yy.cicg.rpi.edu
