Rayshade file formats
Original by Craig Kolb
Edited by Paul Bourke
The Input File
The scene description read by rayshade consists of a number of keywords,
each followed by a set of arguments. These keywords can be thought of as
commands that direct rayshade to do various things, such as create objects,
set the eye's position, and change an object's appearance.
Many of the keywords have related command-line options for turning on
special features and setting values. These options override the values
given in the input file, and are explained in detail later.
Rayshade is "case sensitive," which means that typing SPHERE or Sphere
instead of sphere won't work. Rayshade keywords are all lower-case. Many
people choose to capitalize the first letter of names that they give to
objects or surfaces in order to make then "stand out" in an input file.
Keywords, numbers and strings in the input file are separated by spaces,
tabs, or new lines (carriage returns). These "whitespace" characters are
handled identically by rayshade, which means that you can separate keywords
from keywords, key words from arguments, and arguments from arguments
using any combination of whitespace characters that you choose.
Numbers may be entered directly as reals or as parenthesized expressions.
Reals may be written in exponential notation if you wish, and integers may
be written with or without a trailing decimal point. When an integer value
is called for and a real is given, the real value is truncated and the
resulting integer is used. Table 2.1 lists the available operators
available for use in expressions, in order of decreasing precedence.
The upshot of all this is that the strings 42, 42:, (20 + 22), and (72 - 7)
mean the same thing to rayshade.
Variables may also be defined and used in expressions. Several built-in
functions are also provided, see later for more details.
Rayshade will automatically run the input it receives through the C
preprocessor if it is available. This allows you to use C preprocessor
directives, such as #include, #define, and #ifdef when designing your
Comments may be included in rayshade input files by enclosing text
between the strings /* and */, as in the C programming language.
The end result of running rayshade is an image file. Depending upon how it
was installed, rayshade writes images in either the Utah Raster RLE format
or a generic but easily-manipulated mtv format used by Mark VandeWettering
in his mtv ray tracer. The mtv format consists of a header giving
the resolution of the image followed by interleaved red-green-blue values for
each pixel. The RLE format supports an arbitrary number of color channels,
an alpha channel, comments, a history field, and the ability to treat images
as windows into a larger image. As a result of this flexibility, a number of
rayshade's features are not supported if the mtv format is being used. You
are thus strongly encouraged to obtain a copy of the Utah Raster Toolkit.
If the mtv format is used, the image will (and must) consist of three
eight-bit color channels. If the RLE format is used, the image file consists
of three eight-bit color channels plus an eight-bit alpha channel. Rayshade
also documents in the RLE header the command line and gamma value used
in creating the image.
If more than one frame is rendered, the resulting images are appended
in turn to the image file. The various utilities provided by the Utah Raster
Toolkit can be used to manipulate the resulting "movie" files. If the Toolkit
is not being used, you will probably need to write utility programs to handle
the decidedly non-standard mult-image "mtv" format files.
By default, rayshade writes the computed image to the standard output.
The image file may be written to a file instead by specifying a file name in
the input stream.
Write the computed image to the named file.
The output file name may also be specified on the command line by using
the -O option.
The size of the output image is measured in pixels. The x size is the
number of pixels left-to-right, while the y size is the number of pixels
screen xsize ysize
Use a screen xsize pixels wide by ysize pixels high.
The screen size may also be set on the command line through the -R option.
The default screen size is 512 by 512 pixels.
When rendering an image, it is often advantageous to split the image
into a number of disjoint windows, each of which is rendered on a different
machine. One then combines the images corresponding to the windows into
a final image.
window minx maxx miny maxy
Render the image in the given screen subwindow, specified in
The window must be properly contained within the screen, i.e., minx and
miny must be greater than or equal to zero, while maxx and maxy must
be less than or equal to one. The mtv image file format does not support
windows. The Utah Raster tool rlecomp is useful for reconstructing the full
image from sub-images. By default, the window is equal to (0:; 1:; 0; 1), or
the entire screen.
Gamma correction may also be applied to the three output color channels.
See later for more details.
As it is working, rayshade informs you of its progress by writing messages to
a "report file". By default, the report file is the standard error. The report
itself consists of a number of progress report lines consisting of the number
of eye rays traced, the total elapsed time, and the elapsed time since the
last progress report. The end of the report contains detailed statistics about
intersection tests performed, the number of rays traced, and the like.
report [verbose] [quiet] [freq] [file]
Specify the kind of information included in the report and to which
file it should be written. If verbose is specified, the report will also
include a listing of the options selected, the bounding volumes of each
aggregate, and the total number of primitives in each aggregate. quiet
causes warning messages to be suppressed. freq specifies the frequency,
in scanlines rendered, that progress reports are made. If given, file
names a file to which the report will be written.
By default, a non-verbose, non-quiet report is written to the standard error
once every 10 lines. The -V option may also be used to direct the report to
a named file.
Given a screen of a fixed size, creating an image is accomplished by
sampling each pixel one or more times in order to determine what can be
seen "through" that pixel by the camera. A pixel thus covers a square
area of the image plane, not just a single point.
If a pixel is not sampled at the proper rate, aliasing will result. Aliasing
usually appears as "jaggies" or "stair steps" in the image. In order to reduce
these and other artifacts, rayshade provides an adaptive jittered antialiasing
scheme that attempts to detect where increased sampling rates are needed.
In jittered sampling, the location at which a sample is taken is perturbed by
a random amount. This perturbation reduces aliasing but adds noise to the
The adaptive sampling scheme implemented in rayshade begins by sampling
each pixel on the current scanline once. For each pixel on the scanline,
the contrast between it and its four immediate neighbors is computed. If
this contrast is greater than a user-specified maximum in any color channel,
the pixel and its neighbors are all supersampled by firing an additional
numsamples2 - 1 rays through those pixels that have not already been
supersampled. This process is repeated for the current scanline until
a pass is made without any pixel being supersampled.
contrast redcont greencont bluecont
Set the maximum allowed contrast between four color samples when
adaptive supersampling is used. The contrast test is applied to each
color channel separately.
The default maximum contrast values for the red, green, and blue channels
are 0.25, 0.2, and 0.4, respectively.
sample n [nojitter]
Use n2 samples when performing jittered sampling. The maximum
legal value is 5. If nojitter is specified, sample locations and times
will not be jittered.
By default, 32 jittered samples are taken.
A given set of sample values must be filtered in order to assign a color
to a pixel. Ideally, when performing filtering for a specific pixel,
the filter will consider samples from neighboring regions. In rayshade,
the filtering applied to a pixel makes use of samples taken for that
pixel alone. However, one may increase the size of the filter that is
applied in order to approximate the results a more robust filtering scheme.
filter type [width]
Use the indicated filter type with the given width, in pixels.
Supported filter types are gauss (gaussian) and box (the default).
The default filter width is 1.0 for a box filter, 1.8 for a gaussian
filter. The filter and pixel centers always coincide. When sampling
a pixel, samples are taken over the area of the pixel filter, which
is not necessarily the same as the area of the pixel itself.
Jittered sampling is used in rayshade to sample extended light sources
as well. A total of samples2 samples are taken of each extended light
source in order to determine the extent of shadowing.
The Ray Tree
When ray tracing a scene, reflected or transmitted rays may strike
other reflective or transparent objects. Further reflected or transmitted
rays will be spawned, and so on. Taken together, such a family of rays
is termed the ray tree. Care must be taken to control the depth of
this tree: If it is allowed to grow too deeply, one may spend a great
deal of time computing rays that contribute little to the final picture;
if it is not allowed to grow far enough, this premature tree pruning
may be evident in the image.
Rayshade provides two complimentary methods for controlling the depth
of the ray tree. One method sets an absolute maximum for the tree. The
other allows one to adaptively prune a tree as it grows so that
"unimportant" rays are not spawned.
Do not spawn rays deeper than those at the given level.
Rays from the eye are of depth zero. The default value for level is 15.
This depth may also be set from the command line through the -D option.
Do not spawn rays whose contribution to the final color of the eye ray
is less than threshold for each color channel. Threshold may be given
as a single floating-point value, or as a red-green-blue triple.
The default value is 0.002. This threshold may also be set from the command
line through the -T option.
Specifying a View
When designing a rayshade input file, there are two main issues that must
be considered. The first and more complex is the selection of the objects
to be rendered and the appearances they should be assigned. The second
and usually easier issue is the choice of viewing parameters.
Rayshade uses a camera model to describe the geometric relationship
between the objects to be rendered and the image that is produced. This
relationship describes a perspective projection from world space onto the
The geometry of the perspective projection may be thought of as an
infinite pyramid, known as the viewing frustum. The apex of the frustum
is defined by the camera's position, and the main axis of the frustum by
a "look" vector. The four sides of the pyramid are differentiated by
their relationship to a reference "up" vector from the camera's position.
The image ultimately produced by rayshade may then be thought of as
the projection of the objects closest to the eye onto a rectangular screen
formed by the intersection of the pyramid with a plane orthogonal to the
pyramid's axis. The overall shape of the frustum (the lengths of the top and
bottom sides compared to left and right) is described by the horizontal and
vertical fields of view.
The three basic camera properties are its position, the direction in which it
is pointing, and its orientation. The keywords for specifying these values are
described below. The default values are designed to provide a reasonable
view of a sphere or radius 2 located at origin. If these default values are
used, the origin is projected onto the center of the image plane, with the
world x axis running left-to-right, the z axis bottom-to-top, and the y axis
going "into" the screen.
Place the virtual camera at the given position.
The default camera position is (0, -8, 0).
Point the virtual camera toward the given position.
The default look point is the origin (0, 0, 0). The look point and camera
position must not be coincident.
The "up" vector from the camera point is set to the given direction.
This up vector need not be orthogonal to the view vector, nor need it be
normalized. The default up direction is (0, 0, 1).
Another popular standard viewing geometry, with the x axis running
left-to-right, the y axis bottom-to-top, and the z axis pointing out of the
screen, may be obtained by setting the up vector to (0, 1, 0) and by placing
the camera on the positive z axis.
Field of View
Another important choice to be made is that of the field of view of the
camera. The size of this field describes the angles between the left
and right sides and top and bottom sides of the frustum.
fov hfov [vfov]
Specify the horizontal and vertical field of view, in degrees.
The default horizontal field of view is 45 degrees. If vfov is omitted,
as is the general practice, the vertical field of view is computed using
the horizontal field of view, the output image resolution, and the
assumption that a pixel samples a square area. Thus, the values passed
via the screen keyword define the shape of the final image. If you are
displaying on a non-square pixeled device, you must set the vertical
field of view to compensate for the "squashing" that will result.
Depth of Field
Under many circumstances, it is desirable to render objects in the image
such that they are in sharp focus on the image plane. This is achieved by
using the default "pinhole' camera. In this mode, the camera's aperture is
a single point, and all light rays are focused on the image plane.
Alternatively, one may widen the aperture in order to simulate depth
of field. In this case, rays are cast from various places on the aperture
disk towards a point whose distance from the camera is equal to the focus
distance. Objects that lay in the focal plane will be in sharp focus.
The farther an object is from the image plane, the more out-of-focus it
will appear to be. A wider aperture will lead to a greater blurring of
objects that do not lay in the focal plane. When using a non-zero
aperture radius, it is best to use jittered sampling in order to reduce
Use an aperture with the given radius.
The default radius is zero, resulting in a pinhole camera model.
Set the focal plane to be distance units from the camera.
By default, the focal distance is equal to the distance from the camera to
the look point.
Producing a stereo pair is a relatively simple process; rather than simply
rendering a single image, one creates two related images which may then
be viewed on a stereo monitor, in a stereo slide viewer, or by using colored
glasses and an appropriate display filter.
Rayshade facilitates the rendering of stereo pairs by allowing you to
specify the distance between the camera positions used in creating the two
images. The camera position given in the rayshade input file defines the
midpoint between the two camera positions used to generate the images.
Generally, the remainder of the viewing parameters are kept constant.
Specifies the camera separation to be used in rendering stereo pairs.
There is no default value. The separation may also be specified on the
command line through the -E option. The view to be rendered (left or
right) must be specified on the command line by using the -l or -r options.
There are several things to keep in mind when generating stereo pairs.
Firstly, those objects that lie in from of the focal plane will appear
to protrude from the screen when viewed in stereo, while objects farther
than the focal plane will recede into the screen. As it is usually
easier to look at stereo images that recede into the screen, you will
usually want to place the look point closer to the camera than the object
of primary interest.
The degree of stereo effect is a function of the camera separation and
the distance from the camera to the look point. Too large a separation will
result in a hyperstereo effect that will be hard to resolve, while too little a
value will result in no stereo effect at all. A separation equal to one tenth
the distance from the camera to the look point is often a good choice.
The lighting in a scene is determined by the number, type, and nature of
the light sources defined in the input file. Available light sources range
from simple directional sources to more realistic but computationally costly
quadrilateral area light sources. Typically, you will want to use point or
directional light sources while developing images. When final renderings are
made, these simple light sources may be replaced by the more complex ones.
No matter what kind of light source you use, you will need to specify
its intensity. An Intensity is either a red-green-blue triple indicating
the color of the light source, or a single value that is interpreted as
the intensity of a "white" light. In the current version of rayshade, the
intensity of a light does not decrease as one moves farther from it.
If you do not define a light source, rayshade will create a directional light
source of intensity 1.0 defined by the vector (1, -1, 1). This default light
source is designed to work well when default viewing parameters and surface
values are being used.
You may define any number of light sources, but keep in mind that it
will require more time to render images that include many light sources.
It should also be noted that the light sources themselves will not appear
in the image, even if they are placed in frame.
Light Source Types
The amount of ambient light present in a scene is controlled by a pseudo
light source of type ambient.
light Intensity ambient
Define the amount of ambient light present in the entire scene.
There is only one ambient light source; its default intensity is 1, 1, 1.
If more than one ambient light source is defined, only the last instance is
used. A surface's ambient color is multiplied by the intensity of the ambient
source to give the total ambient light reflected from the surface.
Directional sources are described by a direction alone, and are useful
for modeling light sources that are effectively infinitely far away from
the objects they illuminate.
light Intensity directional direction
Define a light source with the given intensity that is defined to be in
the given direction from every point it illuminates. The direction need
not be normalized.
Point sources are defined as a single point in space. They produce
shadows with sharp edges and are a good replacement for extended and
other computationally expensive light source.
light Intensity point pos
Place a point light source with the given intensity at the given position.
Spotlights are useful for creating dramatic localized lighting effects.
They are defined by their position, the direction in which they are
pointing, and the width of the beam of light they produce.
light Intensity spot pos to [in out]
Place a spotlight at pos, oriented as to be pointing at to. The intensity
of the light falls off as (cosine angle), where angle is the angle between
the spotlight's main axis and the vector from the spotlight to the point
being illuminated. in and out may be used to control the radius of
the cone of light produced by the spotlight.
in is the the angle at which the light source begins to be attenuated. At
out, the spotlight intensity is zero. This affords control over how "fuzzy"
the edges of the spotlight are. If neither angle is given, they both are
effectively set to 180 degrees.
Extended sources are meant to model spherical light sources. Unlike
point sources, extended sources actually possess a radius, and as such are
capable or producing shadows with fuzzy edges (penumbrae). If you do not
specifically desire penumbrae in your image, use a point source instead.
light Intensity extended radius pos
Create an extended light source at the given position and with the
The shadows cast by extended sources are modeled by taking samples of
the source at different locations on its surface. When the source is partially
hidden from a given point in space, that point is in partial shadow with
respect to the extended source, and the sampling process is usually able to
determine this fact.
Quadrilateral light sources are computationally more expensive than extended
light sources, but are more flexible and produce more realistic results.
This is due to the fact that an area source is approximated by a number of
point sources whose positions are jittered to reduce aliasing. Because each
of these point sources has shading calculations performed individually, area
sources may be placed relatively close to the objects it illuminates, and a
reasonable image will result.
light Intensity area p1 p2 usamp p3 vsamp
Create a quadrilateral area light source. The u axis is defined by the
vector from p1 to p2 . Along this axis a total of usamp samples will
be taken. The v axis of the light source is defined by the vector from
p1 to p3. Along this axis a total of vsamp samples will be taken.
The values of usamp and vsamp are usually chosen to be proportional to the
lengths of the u and v axes. Choosing a relatively high number of samples
will result in a good approximation to a "real" quadrilateral source. However,
because complete lighting calculations are performed for each sample, the
computational cost is directly proportional to the product of usamp and vsamp.
In order to determine the color of a point on the surface of any object, it
is necessary to determine if that point is in shadow with respect to each
defined light source. If the point is totally in shadow with respect to a light
source, then the light source makes no contribution to the point's final color.
This shadowing determination is made by tracing rays from the point
of intersection to each light source. These "shadow feeler" rays can add
substantially to the overall rendering time. This is especially true if extended
or area light sources are used. If at any point you wish to disable shadow
determination on a global scale, there is a command-line option (-n) that
allows you to do so. It is also possible to disable the casting of shadows
onto given objects through the use of the noshadow keyword in surface
descriptions. In addition, the noshadow keyword may be given following the
definition of a light source, causing the light source to cast no shadows onto
Determining if a point is in shadow with respect to a light source is
relatively simple if all the objects in a scene are opaque. In this case, one
simply traces a ray from the point to the light source. If the ray hits an
object before it reaches the light source, then the point is in shadow.
Shadow determination becomes more complicated if there are one or
more objects with non-zero transparency between the point and the light
source. Transparent objects may not completely block the light from a
source, but merely attenuate it. In such cases, it is necessary to compute
the amount of attenuation at each intersection and to continue the shadow
ray until it either reaches the light source or until the light is completely
By default, rayshadecomputes shadow attenuation by assuming that the
index of refraction of the transparent object is the same as that of the
medium through which the ray is traveling. To disable partial shadowing
due to transparent objects, the shadowtransp keyword should be given
somewhere in the input file.
The intensity of light striking a point is not affected by intervening
If you enclose an object behind a transparent surface, and you wish the inner
object to be illuminated, you must not use the shadowtransp keyword or
the -o option.
Objects in rayshade are composed of relatively simple primitive objects.
These primitives may be used by themselves, or they may be combined to
form more complex objects known as aggregates. A special family of
aggregate objects, Constructive Solid Geometry or CSG objects, are the
result of a boolean operations applied to primitive, aggregate, or CSG objects.
An instance is an object that has optionally been transformed and textured.
They are the entities that are actually rendered by rayshade; when you
specify that, for example, a textured sphere is to be rendered, you are
said to be instantiating the textured sphere. An instance is specified
as a primitive, aggregate, or CSG object that is followed by optional
transformation and texturing information.
The World Object
Writing a rayshade input file is principally a matter of defining a special
aggregate object, the World object, which is a list of the objects in the
scene. When writing a rayshade input file, all objects that are instantiated
outside of object-definition blocks are added to the World object; you need
not (not should you) define the World object explicitly in the input file.
Primitive objects are the building box with which other objects are created.
Each primitive type has associated with it specialized methods for creation,
intersection with a ray, bounding box calculation, surface normal calculation,
ray enter/exit classification, and for the computation 2D texture coordinates
termed u-v coordinates. This latter method is often referred to as the inverse
While most of these methods should be of little concern to you, the inverse
mapping methods will affect the way in which certain textures are
applied to primitives. Inverse mapping is a matter of computing normalized
u and v coordinates for a given point on the surface of the primitive. For
planar objects, the u and v coordinates of a point are computed by linear
interpolation based upon the u and v coordinates assigned to vertices or other
known points on the primitive. For non-planar objects, uv computation can
be considerably more involved.
This section briefly describes each primitive and the syntax that should
be used to create an instance of the primitive. It also describes the inverse
mapping method, if any, for each type.
blob thresh st r p [st r p . .].
Defines a blob with consisting of a threshold equal to thresh, and a
group of one or more metaballs. Each metaball is defined by its
position p, radius r, and strength st.
For now, see the source code for more explicit documentation. There is no
inverse mapping method for blobs.
box corner1 corner2
Creates an axis-aligned box which has corner1and corner2as opposite
Transformations may be applied to the box if a non-axis-aligned instance is
required. There is no inverse mapping method for boxes.
sphere radius center
Creates a sphere with the given radius and centered at the given position.
Note that ellipsoids may be created by applying the proper scaling to a
sphere. Inverse mapping on the sphere is accomplished by computing the
longitude and latitude of the point on the sphere, with the u value
corresponding to longitude and v to latitude. On an untransformed
sphere, the z axis defines the poles, and the x axis intersects the
sphere at u = 0, v = 0:5. There are degeneracies at the poles:
the south pole contains all points of latitude 0., the north all points
of latitude 1.
torus rmajor rminor center up
Creates a torus centered at center by rotating a circle with the given
minor radius around the center point at a distance equal to the major
In tori inverse mapping, the u value is computed using the angle of rotation
about the up vector, and the v value is computing the angle of rotation
around the tube, with v = 0 occuring on the innermost point of the tube.
triangle p1 p2 p3
Creates a triangle with the given vertices.
triangle p1 n1 p2 n2 p3 n3
Creates a Phong-shaded triangle with the given vertices and vertex normals.
For both Phong- and flat-shaded triangles, the u axis is the vector from p1
to p2 , and the v axis the vector from p1 to p3 . There is a degeneracy at
p3, which contains all points with v = 1:0. This default mapping may be
modified using the triangleuv primitive described below.
triangleuv p1 n1 uv1 p2 n2 uv2 p3 n3 uv3
Creates a Phong-shaded triangle with the given vertices, vertex
normals. When performing texturing, the uv given for each vertex are
used instead of the default values.
When computing uv coordinates within the interior of the triangle, linear
interpolation of the coordinates associated with each triangle vertex is used.
poly p1 p2 p3 [p4 . .].
Creates a polygon with the given vertices. The vertices should be
given in counter-clockwise order as one is looking at the "top" side of
the polygon. The number of vertices in a polygon is limited only by
Inverse mapping for arbitrary polygons is problematical. Rayshade punts
and equate u with the x coordinate of the point of intersection, and v with
the y coordinate.
Creates a height field defined by the altitude data stored in the named
file. The height field is based upon perturbations of the unit square
in the z = 0 plane, and is rendered as a surface tessellated by right
Height field inverse mapping is straight-forward: u is the x coordinate
of the point of intersection, v the y coordinate.
Creates a plane that passes through the given point and has the specified
Inverse mapping on the plane is identical to polygonal inverse mapping.
cylinder radius bottom top
Creates a cylinder that extends from bottom to top and has the
indicated radius. Cylinders are rendered without endcaps.
The cylinder's axis defines the v axis. The u axis wraps around the cylinder,
with u = 0 dependent upon the orientation of the cylinder.
cone radbottombottom radtoptop
Creats a (truncated) cone that extends from bottom to top. The cone
will have a radius of radbottomat bottom and a radius of radtopat top.
Cones are rendered without endcaps.
Cone inverse mapping is analogous to cylinder mapping.
disc radius position normal
Creates a disc centered at the given position and with the indicated
Discs are useful for placing endcaps on cylinders and cones. Inverse mapping
for the disc is based on the computation of the normalized polar coordinates
of the point of intersection. The normalized radius of the point of
intersection is assigned to u, while the normalized angle from a reference
vector is assigned to v.
An aggregate is a collection of primitives, aggregate, and CSG objects. An
aggregate, once defined, may be instantiated at will, which means that copies
that are optionally transformed and textured may be made. If a scene calls
for the presence of many geometrically identical objects, only one such object
need be defined; the one defined object may then be instantiated many times.
An aggregate is one of several possible types. These aggregate types
are differentiated by the type of ray/aggregate intersection algorithm (often
termed an acceleration technique or efficiency scheme) that is used.
Aggregates are defined by giving a keyword that defines the type of the
aggregate, followed by a series of object instantiations and surface
definitions, and terminated using the end keyword. If a defined object
contains no instantiations, a warning message is printed.
The most basic type of aggregate, the list, performs intersection testing
in the simplest possible way: Each object in the list is tested for
intersection with the ray in turn, and the closest intersection is returned.
list . .e.nd
Create a List object containing those objects instantiated between the
The grid aggregate divides the region of space it occupies into a number
of discrete box-shaped voxels. Each of these voxels contains a list of the
objects that intersect the voxel. This discretization makes it possible to
restrict the objects tested for intersection to those that are likely to
hit the ray, and to test the objects in nearly "closest-first" order.
grid xvox yvox zvox . .e.nd
Create a Grid objects composed of xvox by yvox by zvox voxels
containing those objects instantiated between the grid/end pair.
It is usually only worthwhile to "engrid" rather large, complex collections
of objects. Grids also use a great deal more memory than List objects.
Constructive Solid Geometry
Constructive Solid Geometry is the process of building solid objects from
other solids. The three CSG operators are Union, Intersection, and
Difference. Each operator acts upon two objects and produces a single object
result. By combining multiple levels of CSG operators, complex objects can
be produced from simple primitives.
The union of two objects results in an object that encloses the space
occupied by the two given objects. Intersection results in an object that
encloses the space where the two given objects overlap. Difference is an
order dependent operator; it results in the first given object minus the space
where the second intersected the first.
CSG in Rayshade
CSG in rayshadewill generally operate properly when applied to conjunction
with on boxes, spheres, tori, and blobs. These primitives are by nature
consistent, as they all enclose a portion of space (no hole from the "inside"
to the "outside"), have surface normals which point outward (they are not
"inside-out"), and do not have any extraneous surfaces.
CSG objects may also be constructed from aggregate objects. These
aggregates contain whatever is listed inside, and may therefore be
inconsistent. For example, an object which contains a single triangle
will not produce correct results in CSG models, because the triangle
does not enclose space. However, a collection of four triangles which
form a pyramid does enclose space, and if the triangle normals are
oriented correctly, the CSG operators should work correctly on the pyramid.
CSG objects are specified by surrounding the objects upon which to
operate, as well as any associated surface-binding commands, by the operator
verb on one side and the end keyword on the other:
union Object Object [Object . .].end
Specify a new object defined as the union of the given objects.
difference Object Object [Object . .].end
Specify a new object defined as the difference of the given objects.
intersect Object Object [Object . .].end
Specify a new object defined as the intersection of the given objects.
Note that the current implementation does not support more that two
objects in a CSG list (but it is planned for a future version).
Potential CSG Problems
A consistent CSG model is one which is made up of solid objects with no
dangling surfaces. In rayshade, it is quite easy to construct inconsistent
models, which will usually appear incorrect in the final images. In rayshade,
CSG is implemented by maintaining the tree structure of the CSG
operations. This tree is traversed, and the operators therein applied,
on a per-ray basis. It is therefore difficult to verify the consistency
of the model "on the fly."
One class of CSG problems occur when surfaces of objects being operated
upon coincide. For example, when subtracting a box from another box to
make a square cup, the result will be wrong if the tops of the two boxes
coincide. To correct this, the inner box should be made slightly taller than
the outer box. A related problem that must be avoided occurs when two
coincident surfaces are assigned different surface properties.
It may seem that the union operator is unnecessary, since listing two
objects together in an aggregate results in an image that appears to be the
same. While the result of such a short-cut may appear the same on the
exterior, the interior of the resulting object will contain extraneous surfaces.
The following examples show this quite clearly.
box -2 0 -3 2 3 3
union /* change to list; note bad internal surfaces */
sphere 2 1 0 0
sphere 2 -1 0 0
end rotate 1 0 0 -40 rotate 0 0 1 50
The visual evidence of an inconsistent CSG object varies depending upon
the operator being used. When subtracting a consistent object from and
inconsistent one, the resulting object will appear to be the union of the two
objects, but the shading will be incorrect. It will appear to be inside-out
in places, while correct in other places. The inside-out sections indicate the
areas where the problems occur. Such problems are often caused by
polygons with incorrectly specified normals, or by surfaces that exactly
coincide (which appear as partial "swissh-cheese" objects).
The following example illustrates an attempt to subtract a sphere from a
pyramid defined using an incorrectly facing triangle. Note that the resulting
image obviously points to which triangle is reversed.
name pyramid list
triangle 1 0 0 0 1 0 0 0 1
triangle 1 0 0 0 0 0 0 1 0
triangle 0 1 0 0 0 0 0 0 1
triangle 0 0 1 1 0 0 0 0 0 /* wrong order */
object pyramid scale 3 3 3 rotate 0 0 1 45
rotate 1 0 0 -30 translate 0 -3.5 0
sphere 2.4 0 0 0
By default, cylinders and cones do not have end caps, and thus are not
consistent primitives. One must usually add endcaps by listing the cylinder
or cone with (correctly-oriented) endcap discs in an aggregate.
A name may be associated with any primitive, aggregate, or CS object
through the use of the name keyword:
name objname Instance
Associate objname with the given object. The specified object is not
actually instantiated; it is only stored under the given name.
An object thus named may then be instantiated (with possible additional
transforming and texturing) via the object keyword:
object objname [Transformations] [Textures]
Instantiate a copy of the object associated with objname. If given, the
transformations and textures are composed with any already associated
with the object being instantiated.
Surfaces and Atmospheric Effects
Surfaces are used to control the interaction between light sources and
objects. A surface specification consists of information about how the light
interacts with both the exterior and interior of an object . For non-closed
objects, such as polygons, the "interior" of an object is the "other side" of
the object's surface relative to the origin of a ray.
Rayshade usually ensures that a primitive's surface normal is pointing
towards the origin of the incident ray when performing shading calculations.
Exceptions to this rule are transparent primitives, for which rayshade uses
the direction of the surface normal to determine if the incident ray is
entering or exiting the object. All non-transparent primitives will,
in effect, be double-sided.
A surface definition consists of a number of component keywords, each of
which is usually followed by either a single number or a red-green-blue color
triple. Each of the values in the color triple are normalized, with zero
indicating zero intensity, and one indicating full intensity.
If any surface component is left unspecified, its value defaults to zero,
with the exception of the index of refraction, which is assigned the default
index of refraction (normally 1.0).
Surface descriptions are used in rayshade to compute the color of a ray
that strikes the surface at a point P . The normal to the surface at P, N,
is also computed.
Use the given color to approximate those surface-surface interactions
(eg, diffuse interreflection) not modeled by the ray tracing process.
A surface's ambient color is always applied to a ray. The color applied is
computed by multiplying the ambient color by the intensity of the ambient
If P is in shadow with respect to a given light source, that light source
makes no contribution to the shading of P .
Specifies the diffuse color.
The diffuse contribution from each non-shadowed light source at P is equal
to the diffuse color of the surface scaled by the cosine of the angle between
N and the vector from P to the light source.
Specifies the base color of specular reflections.
Specifies the specular highlight) exponent.
The intensity of specular highlights from light sources are scaled by the
specular color of the surface.
Specifies the specular reflectivity of the surface. If non-zero, reflected
rays will be spawned.
The intensity of specularly reflected rays will be proportional to the
specular color of the surface scaled by the reflectivity.
Specifies the specular transmissivity of the surface. If non-zero,
transmitted (refracted) rays will be spawned.
Specifies the body color of the object. The body color affects the color
of rays that are transmitted through the object.
Specifies the extinction coefficient of the interior of the object.
The extinction coefficient is raised to a power equal to the distance the
transmitted ray travels through the object. The overall intensity of
specularly transmitted rays will be proportional to this factor multiplied
by the surface's body color multiplied by the transparency of the object.
Specifies the index of refraction. The default value is equal to the
index of refraction of the atmosphere surrounding the eye.
translucency translu colorstexp
Specifies the translucency, diffusely transmitted color, and Phong
exponent for transmitted specular highlights.
If a light source illuminates a translucent surface from the side opposite
that from which a ray approaches, illumination computations are performed,
using the given color as the surface's diffuse color, and the given exponent
as the Phong highlight exponent. The resulting color is then scaled by the
Any number of atmospheric effects may also be associated with a surface.
These effects will be applied to those rays that are transmitted through the
surface. Applying atmospheric effects to opaque objects is a waste of input
Add exponential fog with the specified thinness and color.
Fog is simulated by blending the color of the fog with the color of each ray.
The amount of fog color blended into a ray color is an exponential function
of the distance from the ray origin to the point of intersection divided by the
specified thinness for each color channel. If the distance is equal to thinness,
a ray's new color will be half of the fog color plus half its original color.
mist colorthinness zero scale
Add global low-altitude mist of the specified color. The color of a
ray is modulated by a fog with density that varies linearly with the
difference in z coordinate (altitude) between the ray origin and the
point of intersection. The thinness values specify the transmissivity of
the fog for each color channel. The base altitude of the mist is given
by zero, and the apparent height of the mist can be modulated using
scale, which scales the difference in altitude used to compute the fog.
The Default Medium
The default medium is the medium which surrounds and encompasses all
of the objects in the scene; it is the "air" through which eye rays usually
travel before hitting an object. The properties of the default medium may
be modified through the use of the atmosphere keyword.
atmosphere [N ] [atmospheric effects]
If given, N specifies the index of refraction of the default medium. The
default is 1.0. Any atmospheric effects listed are applied to rays that
are exterior to every object in the scene (e.g., rays emanating from the
* Red sphere on a grey plane, with fog.
eyep 0. -10. 2.
atmosphere fog .8 .8 .8 14. 14. 14.
plane 0 0 0 0 0 1
sphere diffuse 0.8 0 0 1.5 0 0 1.5
Rayshade provides a number of ways to define surfaces and to bind these
surfaces to objects. The most straight-forward method of surface
specification is to simply list the surface properties to be used.
Alternatively, one may associate a name with a given surface. This
name may subsequently be used to refer to that surface.
surface name SurfaceDefinition
Associate the given collection of surface attributes with the given
The binding of a collection of surface properties to a given object is
accomplished in a bottom-up manner; the surface that "closest" in the
modeling tree to the primitive being rendered is the one that is used
to give the primitive its appearance.
An object that has no surface bound to it is assigned a default surface
that give an object the appearance of white plastic.
The first and most direct way to bind a surface to a primitive is by
specifying the surface to be bound to the primitive when it is instantiated.
This is accomplished by inserting a list of surface attributes or a surface
name after the primitive's type keyword and before the actual primitive
* A red 'mud' colored sphere reseting on a
* white sphere. To the right is a sphere with
* default surface attributes.
surface mud ambient .03 0. 0. diffuse .7 .3 0.
sphere ambient .05 .05 .05 diffuse .7 .7 .7 1. 0 0 0
sphere mud 1. 0 0 2
sphere 1. 1.5 0 0
Here, we define a red surface named "mud". We then instantiate a
sphere, which has a diffuse white surface bound to it. The next line
instantiates a sphere with the defined "mud" surface bound to it.
The last line instantiates a sphere with no surface bound to it; it
is assigned the default surface by rayshade.
The applysurf keyword may be used to set the default surface
characteristics for the aggregate object currently being defined.
The specified surface is applied to all following instantiated objects
that do not have surfaces associated with them. The scope of this
keyword is limited to the aggregate currently being defined.
* Mirrored ball and cylinder sitting on 'default' plane.
surface mirror .01 .01 .01 diffuse .05 .05 .05
specular .8 .8 .8 speccoef 20 reflect 0.95
plane 0 0 0 0 0 1
sphere 1 0 0 0
cylinder 1 3 0 0 3 0 3
For convenience, the name cursurf may be used to refer to the current
The utility of bottom-up binding of surfaces lies in the fact that one
may be as adamant or as noncommittal about surface binding as one sees
fit when defining objects. For example, one could define a king chess piece
consisting of triangles that have no surface bound to them, save for the cross
on top, which has a gold-colored surface associated with it. One may then
instantiate the king twice, once applying a black surface, and once applying
a white surface. The result: a black king and a white king, each adorned
with a golden cross.
surface white ...
surface black ...
surface gold ...
box x y z x y z
triangle x y z x y z x y z
object gold cross
object white king translate 1. 0 0
object black king
Rayshade supports the application of linear transformations to objects and
textures. If more than one transformation is specified, the total resulting
transformation is computed and applied.
Translate (move) by delta.
rotate axis angle
Rotate counter-clockwise about the given axis by angle degrees.
Scale by v
All three scaling components must be non-zero, else degenerate matrices will
transform row1 row2 row3 [delta]
Apply the given 3-by-3 transformation matrix. If given, delta specifies
a translation vector.
Transformations should be specified in the order in which they are to be
applied immediately following the item to be transformed. For example:
* Ellipsoid, rotated cube
sphere 1. 0 0 0 scale 2. 1. 1. translate 0 0 -2.5
box 0 0 0 .5 .5 .5
rotate 0 0 1 45 rotate 1 0 0 45 translate 0 0 2.5
Transformations may also be applied to textures:
plane 0 0 -4 0 0 1
texture checker red scale 2 2 2 rotate 0 0 1 45
Note that transformation parameters may be specified using of animated
expressions, causing the transformations themselves to be animated.
Textures are used to modify the appearance of an object through the use
of procedural functions. A texture may modify any surface characteristic,
such as diffuse color, reflectivity, or transparency, or it may modify the
surface normal ("bump mapping") in order to give the appearance of a
Any number of textures may be associated with an object. If more than
one texture is specified, they are applied in the order given. This allows
one to compose texturing functions and create, for example a tiled marble
ground plane using the checker and marble textures.
Textures are associated with objects by following the object specification
by a number of lines of the form:
texture name TexturingArguments [Transformations]
Transformations may be applied to the texture in order to, for example,
shrink or grow feature size, change the orientation of features, and change
the position of features.
Several of the texturing functions take the name of a colormap as an
argument. A colormap is 256-line ASCII file, with each line containing
three space-separated values ranging from 0 to 255. Each line gives the red,
green, and blue values for a single entry in the colormap.
blotch BlendFactor surface
Produces a mildly interesting blotchy-looking surface. BlendFactor is
used to control the interpolation between the default surface
characteristics and the characteristics of the given surface.
A value of 0 results in a roughly 50-50 mix of the two surfaces.
Higher values result in a great portion of the default surface
Apply a random bump map. The point of intersection is passed to
DNoise(). The returned normalized vector is weighted by scale and
the result is added to the normal vector at the point of intersection.
Applies a 3D checkerboard texture. Every point that falls within an
"even" unit cube will be assigned the characteristics of the named
surface applied to it, while points that fall within "odd" cubes will
have its usual surface characteristics. Be wary of strange effects due
to roundoff error that occur when a planar checkered surface lies in a
plane of constant integral value (e.g., z = 0) in texture space. In such
cases, simply translate the texture to ensure that the planar surface is
not coincident with an integral plane in texture space (e.g., translate
0 0 0.1).
cloud scale H ~ octaves cthresh lthresh tscale
This texture is a variant on Geoff Gardner's ellipsoid-texturing
algorithm. It should be applied to unit spheres centered at the
origin. These spheres may, of course, be transformed at will to form
the appropriately-shaped cloud or tree.
A sample of normalized fBm (see the fbm texture) is generated at
the point of intersection. This sample is used to modulate the surface
transparency. The final transparency if a function of the sample value,
the the proximity of the point of intersection to the edge of the sphere
(as seen from the ray origin), and three parameters to control the
overall "density." The proximity of the point to the sphere edge is
determined by evaluating a limb function, which varies from 0 on the
limb to 1 at the center of the sphere.
fBm - cthresh - (lthresh - cthresh)(1 - limb)
transp = 1: - _____________________________________________
fbm offset scale H ~ octaves thresh [colormap]
Generate a sample of discretized fractional Brownian motion (fBm)
and uses it to scale the diffuse and ambient component of an object's
surface. Scale is used to scale the value returned by the fBm function.
Offset allows one to control the minimum value of the fBm function. H
is the Holder exponent used in the fBm function (a value of 0.5 works
well). ~ is used to control lacunarity, and specifies the the frequency
difference between successive samples of the fBm basis function (a value
of 2.0 will suffice). Octaves specifies the number of octaves (samples)
to take of the fBm basis function (in this case, Noise()). Between five
and seven octaves usually works well. Thresh is used to specify a lower
bound on the output of the fBm function. Any value lower than thresh
is set to zero.
If a colormap is named, a 256-entry colormap is read from the named
file, and the sample of fBm is scaled by 255 and is used as an index
into the colormap. The resulting colormap entry is used to scale the
ambient and diffuse components of the object's surface.
fbmbump offset scale H ~ octaves
Similar to the fbm texture. Rather than modifying the color of a
surface, this texture acts as a bump map.
Gives reflective surfaces a glossy appearance. This texture perturbs
the object's surface normal such that the normal "samples" a cone
of unit height with radius 1: - glossiness. A value of 1 results in
perfect mirror-like reflections, while a value of 0 results in extremely
fuzzy reflections. For best results, jittered sampling should be used to
render scenes that make use of this texture.
Gives a surface a marble-like appearance. The texture is implemented
as roughly parallel alternating veins of marble, each of which is
separated by 1/7 of a unit and runs perpendicular to the Z axis. If
a colormap is named, the surface's ambient and diffuse colors will be
scaled using the RGB values in the colormap. If no colormap is given,
the diffuse and ambient components are simply scaled by the value of
the marble function. One may transform the texture to control the
density and orientation of the marble veins.
sky scale H ~ octaves cthresh ltresh
Similar to the fbm texture. Rather than modifying the color of a
surface, this texture modulates its transparency. cthresh is the value
of the fBm function above which the surface is totally opaque. Below
lthresh, the surface is totally transparent.
stripe Surface size bump Mapping
Apply a "raised" stripe pattern to the surface. The surface properties
used to color the stripe are those of the given surface. The width of
the stripe, as compared to the unit interval, is given by size. The
magnitude of bump controls the extent to which the bump appears to
be displaced from the rest of the surface. If negative, the stripe will
appear to sink into the surface; if negative, it will appear to stand out
of the surface.
Gives a surface a wood-like appearance. The feature size of this texture
is approximately 0:01 of a unit, making it often necessary to scale the
texture in order to achieve the desired appearance.
Rayshade also supports an image texture. This texture allows you to use
images to modify the characteristics of a surface. You can use three-channel
images to modify the any or all of the ambient, diffuse, and specular colors
of a surface. If you are using the Utah Raster Toolkit, you can also use
single-channel images to modify surface reflectance, transparency, and the
specular exponent. You can also use a single-channel image to apply a bump
map to a surface.
In all but the bump-mapping case, a component is modified by multiplying
the given value by the value computed by the texturing function. When
using the Utah Raster Toolkit, surface characteristics are modified in
proportion to the value of the alpha channel in the image. If there is no
alpha channel, or you are not using the Utah Raster Toolkit, alpha is
assumed to be everywhere equal to 1.
The named component will be modified.
Possible surface components are: ambient (modify ambient color), diffuse
(modify diffuse color), specular (modify specular color), specpow, (modify
specular exponent), reflect, (modify reflectivity), transp (modify
transparency), bump, (modify surface normal). The specpow, reflect, transp,
and bump components require the use of a single-channel image.
range high low
Specify the range of values to which the values in the image should be
mapped. An value of 1 will be mapped high, 0 to low. Intermediate
values will be linearly interpolated.
When given, pixel averaging will be performed in order to smooth the
sampled image. If not specified, no averaging will occur.
For use when modifying surface colors, this keyword specifies that the
given surface should be used as the base to be modified when the
alpha value in the image is non-zero. When alpha is zero, the object's
unmodified default surface characteristics are retained.
The usual behavior is for the object's default surface properties to be used.
tile un vn
Specify how the image should be tiled (repeated) along the u and
v axes. If positive, the value of un gives the number of times the
image should be repeated along the u axis, starting from the origin
of the texture, and positive vn gives the number of times it should be
repeated along the v axis. If either value is zero, the image is repeated
infinitely along the appropriate axis.
Tiling is usually only a concern when linear mapping is being used, though
it may also be used if image textures are being scaled. By default un and
vn are both zero.
A mapping function may also be associated with an image texture.
Mapping functions are used to apply two-dimensional textures to surfaces.
Each mapping functions defines a different method of transforming a three
dimensional point of intersection to a two dimensional u - v pair termed
texturing coordinates. Typically, the arguments to a mapping method define
a center of a projection and two non-parallel axes that define a local
The default mapping method is termed u - v mapping or inverse mapping
Normally, there is a different inverse mapping method for each primitive
type. When inverse mapping is used, the point of intersection is passed
to the uv method for the primitive that was hit.
Use the uv (inverse mapping) method associated with the object that
was intersected in order to map from 3D to determine texturing
map linear [origin vaxisuaxis]
Use a linear mapping method. The 2D texture is transformed so that
its u axis is given by uaxis and its v axis by vaxis. The texture is
projected along the vector defined by the cross product of the u and
v axes, with the (0,0) in texture space mapped to origin.
map cylindrical [origin vaxisuaxis]
Use a cylindrical mapping method. The point of intersection is
projected onto an imaginary cylinder, and the location of the projected
point is used to determine the texture coordinates. If given, origin
and vaxisdefine the cylinder's axis, and uaxis defines where u = 0 is
See the description of the inverse mapping method for the cylinder.
By default, the point of intersection is projected onto a cylinder that
runs through the origin along the z axis, with uaxis equal to the x axis.
map spherical [origin vaxisuaxis]
Use a spherical mapping method. The intersection point is projected
onto an imaginary sphere, and the location of the projected point is
used to determine the texturing coordinates in a manner identical to
that used in the inverse mapping method for the sphere primitive. If
given, the center of the projection is origin, vaxis defines the sphere
axis, and the point where the non-parallel uaxis intersects the sphere
defines where u = 0 is located.
By default, a spherical mapping projects points towards the origin, with
vaxisdefined to be the z axis and uaxis defined to be the x axis.
Rayshade provides basic animation animation support by allowing timevarying
transformations to be associated with primitives and aggregate objects
Commands are provided for controlling the amount of time between each frame,
the speed of the camera shutter, and the total number of frames to be rendered.
By default, rayshade renders a single frame, with the shutter open for an
instant (0 units of time, in fact). The shutter speed in no way changes the
light-gathering properties of the camera, i.e. frames rendered using a longer
exposure will not appear brighter than those with a shorter exposure. The
only change will be in the potential amount of movement that the frame
"sees" during the time that the shutter is open.
Each ray cast by rayshade samples a particular moment in time. The
time value assigned to a ray ranges from the starting time of the current
frame to the starting time plus the amount of time the shutter is open.
When a ray encounters an object or texture that possesses an animated
transformation, the transformed entity is moved into whatever position is
appropriate for the ray's current time value before intersection, shading, or
texturing computations are performed.
The starting time of the current frame is computed using the length
of each frame the current frame number, and the starting time of the first
Specifies that the shutter is open for t units of time for each exposure.
A larger value of t will lead to more motion blur in the final image. Note
that t may be greater than the actual length of a frame. By default, t is
zero, which prevents all motion blur.
Specifies the time increment between frames.
The default time between frames is 1 unit.
Specifies the starting time of the first frame.
By default, time is zero.
Variables may be defined thorugh the use of the define keyword:
define name value
Associate name with the given value. Value may be a constant or a
The variable name may thereafter be used in expressions in the input file.
An animated transformation is one for which animated expressions have
been used to define one or more of its parameters (e.g. the angle through
which a rotation occurs). An animated expression is one that makes use of
a time-varying ("animated") variable or function.
There are two supported animated variables. The first, time, is equal
to the current time. When a ray encounters an animated transformation
defined using an expression containing time, the ray substitutes its time
value into the expression before evaluation. Using the time variable in an
animated expression is the most basic way to create blur-causing motion.
The second animated variable, frame, is equal to the current frame number
Unlike the time variable, frame takes on a single value for the duration
of each frame. Thus, transforms animated through the use of the frame
variable will not exhibit motion blurring.
Also supported is the linear function. This function uses time implicitly
to interplate between two values.
linear ( Stime, Sval, Etime, Eval )
Linearly interpolate between Sval at time Stime and Eval at time Etime.
If the current time is less than Stime, the function returns Sval.
If the current time is greater than Etime, Eval is returned.
The following example shows the use of the time variable to animate a
sphere by translating it downwards over five frames. Note thet the shutter
keyword is used to set the shutter duration in order to induce motion
sphere 1 0 0 2 translate 0 0 (-time)
Further examples of animation may be found in the Examples directory
of the rayshade distribution.
Height Field Files
This section describes the format of the files that store data for the height
field primitive. The format is an historical relic; a better format is needed.
Height field data is stored in binary form. The first record in the file is
a 32-bit integer giving the square root of number of data points in the file.
We'll call this number the size of the height field.
The size is followed by altitude (z) values stored as 32-bit floating point
values. The 0th value in the file specifies the z coordinate of the lower
left corner of the height field (0, 0). The next specifies the Z coordinate
for (1=(size - 1); 0). The last specifies the coordinate for (1:; 1:). In other
words, the ith value in the heightfield file specifies the z coordinate for the
point whose x coordinate is (i%size)=(size - 1), and whose y coordinate is
(i=size)=(size - 1). Non-square height fields may be rendered by specifying
altitude values less than or equal to the magic value -1000. Triangles that
have any vertex less than or equal in altitude to this value are not rendered.
While this file format is compact, it sacrifices portability for ease of
use. While creating and handling height field files is simple, transporting
a height field from one machine to another is problematical due to the fact
that differences in byte order and floating-point format between machines is
not taken into account.
These problems could be circumvented by writing the height field file in
a fixed-point format, taking care to write the bytes that encode a given
value in a consistent way from machine to machine. An even better idea
would be to write a set of tools for manipulating arbitrary 2D arrays of
floating-point values in a compact, portable way, allowing for comments
and the like in the file. . .