Version 3.9.

Rendering
A Comprehensive User’s Guide
Rendering: A Comprehensive User’s Guide was written by Edna Kruger;
updated by Maggie Kathwaroon; and edited by Edna Kruger and John Woolfrey.
Layout by Luc Langevin.
Special thanks to Craig Hall and Pierre Tousignant for their assistance in
assuring the technical integrity of this guide.
© 1996–2000 Avid Technology, Inc. All rights reserved.
SOFTIMAGE and Avid are registered trademarks of Avid Technology, Inc. or
its subsidiaries or divisions. mental ray and mental images are registered
trademarks of mental images GmbH & Co. KG in the U.S.A. and/or other
countries. All other trademarks contained herein are the property of their
respective owners.
This document is protected under copyright law. The contents of this document
may not be copied or duplicated in any form, in whole or in part, without the
express written permission of Avid Technology, Inc. This document is supplied
as a guide for the SOFTIMAGE|3D product. Reasonable care has been taken in
preparing the information it contains. However, this document may contain
omissions, technical inaccuracies, or typographical errors. Avid Technology,
Inc. does not accept responsibility of any kind for customers’ losses due to the
use of this document. Product specifications are subject to change without
notice.
Printed in Canada.
1100
 Table of Contents

Contents
C H A P T E R O N E

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Introduction to Rendering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
The Camera . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

C H A P T E R T W O

Rendering Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Previewing before Rendering . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Optimizing Rendering Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Rendering a Sequence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Using the Standalone Renderer . . . . . . . . . . . . . . . . . . . . . . . . . 32
Rendering a Subregion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Showing and Hiding Object Edges . . . . . . . . . . . . . . . . . . . . . . . 39
Viewing Rendered Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

C H A P T E R T H R E E

Advanced Rendering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Raytracing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Antialiasing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Blurring a Moving Object . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Field Rendering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Rendering Tag and Z Channels . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Rendering Faces of an Object . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

C H A P T E R F O U R

Rendering with mental ray Software . . . . . . . . . . . . . . . 65

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Overview for Using mental ray . . . . . . . . . . . . . . . . . . . . . . . . . . 72
Switching Between mental ray versions . . . . . . . . . . . . . . . . . . . 73
Previewing with mental ray . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Disabling Effects When Previewing and Rendering . . . . . . . . . . . 77
Using Shaders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

SOFTIMAGE|3D Rendering 3
Table of Contents

Raytracing Using mental ray . . . . . . . . . . . . . . . . . . . . . . . . . . . .84

Antialiasing Using mental ray . . . . . . . . . . . . . . . . . . . . . . . . . . .88
Blurring a Moving Object Using mental ray . . . . . . . . . . . . . . . . .91
Creating mental ray Shadow Maps . . . . . . . . . . . . . . . . . . . . . . .99
Creating mental ray Displacement Maps . . . . . . . . . . . . . . . . . .103
Rendering Faces of an Object . . . . . . . . . . . . . . . . . . . . . . . . . .106
Rendering Shadows. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .107
Surface Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .108
Using Output Shaders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .112
Saving to a File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .116
Distributed mental ray Rendering
between IRIX and Windows NT . . . . . . . . . . . . . . . . . . . . . . . . .117
Interactive Selection of Hosts. . . . . . . . . . . . . . . . . . . . . . . . . . .120
Enabling mental ray Messages During Rendering. . . . . . . . . . . .122
Rendering a Range of Frames . . . . . . . . . . . . . . . . . . . . . . . . . .123

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .125

4 Rendering SOFTIMAGE|3D
 C H A P T E R O N E

Introduction

SOFTIMAGE|3D Rendering 5
Introduction

6 Rendering SOFTIMAGE|3D
 Introduction to Rendering

Introduction to Rendering
An object in SOFTIMAGE|3D is defined by points in space that are
connected together to create a surface. An important part of the
process of creating a 3D object or scene in SOFTIMAGE|3D is the
material definition of the object’s surface, the lighting of the scene,
and rendering. The Matter module contains most of the tools
designed to accomplish all these tasks.
For the purpose of rendering, SOFTIMAGE|3D divides these
surfaces into triangles, which is known as tessellation. Rendering is an
integral part of finalizing an object’s material definition. When you
render an object, SOFTIMAGE|3D processes the object’s surface
triangles and normals relative to the light source and creates a visible
surface that is shaded according to the parameters set for material
and texture attributes.
Each triangle is a planar surface with its front face oriented in one
direction. The orientation of this surface is shown by a vector
(direction line) called a normal, located on the points (vertices) of
each triangle. The normal always needs to be oriented in the
direction of the camera to make that surface visible (you can,
however, make all surfaces visible to the camera – see Rendering Faces
of an Object on page 64).

Lighting
If you render an object without defining a light source, SOFTIMAGE|3D
automatically creates a light. The default light is white and infinite (it’s
far away like the sun), and does not cast shadows. When you create your
own light source, the default light disappears.
When the surface has been defined, the rendering process calculates
all these attributes to create a final image. SOFTIMAGE|3D calculates
the relation between the orientation of surface normals and the light
source to determine the surface attributes of each triangle. For more
information about light, see the Light chapter in the Defining
Materials and Textures User’s Guide.

SOFTIMAGE|3D Rendering 7
Introduction

The Camera
The camera is the device that lets you view your scene – what the
camera sees is displayed in the Perspective window. Ultimately, the
camera’s view is what is rendered for your final image. By default,
this is whatever is shown in window B.
When rendering with the Depthcue, Hardware, Hidden-Line,
Wireframe, and Ghost, Rotoscope (Wire), or Rotoscope (Shade)
rendering types, you can use any of the displayed windows.
Rendering with the other rendering types must be done in window B
with the Perspective view.
You can see the camera itself in the parallel projection windows by
choosing the Camera > Show command. The camera is represented
by a camera icon. It also has an interest to which the camera always
points. The interest’s icon is three intersecting lines, like a null object.
You can use more than one camera while working on your scene by
displaying more than one Perspective window. This allows you to view
your scene in different ways and help you determine exactly how to
render your scene. These additional cameras are non-rendering.
There are many parameters related to the camera, including its roll,
field of view, and depth of field, as well as the picture format, etc.
Many of these parameters can be animated.
For more information on moving the camera around in a scene, see
Using the Camera on page 46 in the Working with SOFTIMAGE|3D
User’s Guide.

Setting Up and Resetting

The Camera > Settings command lets you set up all parameters for
the camera that renders your scene. These include things such as
position of the camera and its interest, camera angle, depth of field,
lens shaders, etc. To reset the camera and all its parameters to their
default values, choose the Camera > Reset command.
When you choose the Camera > Settings command, the Camera
Settings dialog box opens in which you can set the parameters. Ths
following sections describe some of the parameters that have a large
influence on how a scene is viewed and rendered.

8 Rendering SOFTIMAGE|3D
 The Camera

Changing the Angle

The camera lens allows you to define your angle of view in degrees.
The Automatic setting calculates the camera angle for you depending
on its current attributes. For example, if you zoom in to your scene in
the Perspective window, the camera angle automatically increases.
The Custom Angle option lets you specify the angle of the camera in
degrees. A wider angle lets you see more of the scene.
The camera lens angle is based on a 35mm Cine format. Default
values for the camera angle are based on 1 SOFTIMAGE|3D unit = 1
foot. SOFTIMAGE|3D units are arbitrary units that you can define,
so you can effectively use the camera lens angle values as you like.

Depth of Field
The Depth of Field Simulation options provide you with additional
control over camera settings, as well as support for lens shaders used
when rendering with mental ray. The options define the focus of
objects according to their distance from the camera, similar to the
way a real camera works. Depth of field refers to the minimum and
maximum distance from the camera that objects are in focus.
Objects closer than the minimum distance and farther than the
maximum distance become increasingly out of focus.
Note: The mental ray renderer is not a part of the standard
Note
SOFTIMAGE|3D GT package, but may be purchased
separately.
If you select the Automatic option in the Depth of Field area of the
dialog box, it allows you to set standard camera parameters to
modify depth of field. These three parameters work together to
achieve the desired result: Focal Length, F/stop, and Distance.
• Focal length allows you to define the length of the camera lens. A
larger value increases the lens length and decreases depth of field.
For example, 50mm is the standard lens size of a 35mm camera.
• F/stop allows you to set the size of the aperture opening. A smaller
value results in a larger opening, but decreases the depth of field.
• Distance allows you to define the distance from the camera that
objects are at their sharpest focus. Objects located in front of and
beyond this point become out of focus.

SOFTIMAGE|3D Rendering 9
Introduction

If you select the Custom option, you can custom determine your
own settings. These are the parameters that you can set:
• Near Focus is the nearest distance from the camera where objects
are in focus. This is determined in system units.
• Far Focus is the farthest distance from the camera where objects are
in focus. This is determined in system units.
• Max COC is the Maximum Circle of Confusion (COC). It controls
the out-of-focus effect in terms of pixels. The higher the value, the
more out of focus the object is. The default is 20.
For example, if you preview with a resolution of 100 x 100 and the
Max COC is set to 5, you must compensate during the final render. If
the final render is 1000 x 1000, the Max COC must be increased to
50. The Max COC is proportional to the image resolution.
• Max occurs at is the distance from the camera where Max COC
occurs. This is determined in system units.

Choosing an Aspect Ratio

When you render an image, you can select its aspect ratio of the
image by choosing the Camera > Picture Format command.
It is very important to set the aspect ratio when you first start
working on a production. This ensures that what you see in the
Perspective window accurately shows you what is within the frame in
the final rendering.
When you choose this command, the Picture Format dialog box
appears. You can select standard aspect ratios for film, slides, and
video. To customize a non-standard picture format, specify the
aspect ratio, or define the size of the image (width and height).

Using Lens Shaders

Lens shaders are only available when you are using the mental ray
renderer. You can apply one or more lens shaders to the camera. The
effects that lens shaders can produce can be compared to the lens filters
of a real camera and produce such effects as lens flares or fish-eye lens.
Note: The mental ray renderer is not a part of the standard
Note
SOFTIMAGE|3D GT package, but may be purchased
separately.

10 Rendering SOFTIMAGE|3D
 The Camera

The following is an example of a fish-eye lens.

1. Choose the Camera > Settings command. The Camera Settings
dialog box appears.
2. In the Lens Shader area, click the Select button. This opens the
database browser to the Camera Shaders chapter.
3. Select a camera shader from the list and click Load. The camera
shader appears in the list of lens shaders.
4. Choose Preview > Setup in the Matter module and select mental
ray as the Preview Renderer.
5. Choose Preview > All to preview your scene through the fish eye
lens.

Using the Lens Flare Shader

The following is an example of how to use a lens shader called a Lens
Flare.
1. Create a small city scene with two or three cubes over a plane and
one point light.
2. Animate the camera over 10 frames so that the light disappears
behind one building.
3. Choose Camera > Settings. In the Lens Shaders area of the dialog
box, click the Select button and select the Flares_stars lens shader
from the browser that appears.
4. Edit this shader by clicking the Edit button, clicking Select, and
selecting the light source. Select the Circles option and deselect
Streaks.
5. Select the light source and choose Light > Edit. Give it a raytraced
shadow and change the Umbra value to 0.04. You need to lower
the umbra value to make the light fade when it is behind an
object. An umbra of 0 makes the light fade out completely behind
an object.
6. Choose the Render menu command in the Matter module to
render the animation using mental ray as the rendering type.
For more information, see Rendering with mental ray Software on
page 65.

SOFTIMAGE|3D Rendering 11
Introduction

Controlling the Camera’s Movement

When a camera reaches a vertical axis with its interest, it flips. To prevent
this, you can constrain the y-axis (up vector) of the camera to another
object. There are two reasons to control the up vector of the camera:
• To prevent the camera from flipping when it reaches a vertical axis with
the interest. For example, you can produce a proper rollercoaster-like
action with the camera using an up vector constraint.
• To control the banking of the camera using a separate control path.
To use the up vector constraint in its simplest form, follow these
steps. This uses a camera as the example object:
1. Select the camera.
2. Choose Constraint > Up Vector in the Actor or Motion module and
select the object to which you want to constrain the camera’s up
vector.
Using an independent object is not always the most effective way to
control the up vector. If the up vector constraint coincides with the
interest, a flip occurs on the current vertical axis. This problem can
be solved by creating the following nulls and constraints. This
procedure makes the up vector follow the “point of view” of an
animated object exactly, allowing you to animate the camera the
same way as you would animate any object:
1. Select an object.
2. Create three nulls:
- Place one null in front of the object.
- Place one null in the centre of the object as its “point of view.”
- Place one null on the y-axis directly above the “point of view” null.
3. Make these nulls children of the object so that they follow the
object’s movements.
4. Select the camera and constrain it positionally (Constraint >
Position) to the null in the centre of the object (the “point of view”
null).
5. Select the camera interest and constrain it positionally
(Constraint > Position) to the null in front of the object.
6. Select the camera, choose Constraint > Up Vector, and pick the
null directly above the object’s centre on the y-axis.

12 Rendering SOFTIMAGE|3D
 The Camera

7. Animate the object as you like.

Now when the object moves, the camera’s up vector is oriented
correctly because it is constrained to the null above the object.

Representing an Asymmetrical Camera Projection

The Camera Projection Plane allows for the representation of an off-
axis/asymmetrical camera projection.
The perspective projection is in one of two modes: symmetrical
(normal SOFTIMAGE|3D camera) or asymmetrical (if the
projection plane parameters are activated).
These parameters are accessible only through the Spreadsheet
window. To have them listed in the spreadsheet, do the following:
1. In the Spreadsheet window, click QUERY. The Set Query
dialogue box is displayed.
2. At the bottom of the dialogue box, in the Edit Clause section, set
the following clause:
chptyp == CAM
and click Accept. Be sure that you are editing the main query, do
not add another clause or a subquery
3. Click Columns to open the columns dialogue box, and expand
the camera parameters in Available Columns by double-clicking
on CAMERAS.
4. Scroll to the bottom of the list to see the camera projection plane
attributes. Highlight the following:
- cmpdst (camera projection plane distance)
- cmpszx (camera projection plane size in x)
- cmpszy (camera projection plane size in y)
- cmpofx (camera projection plane offset in x)
- cmpofy (camera projection plane offset in y)
- cmpact (camera projection plane active)
5. Click Add to place them in the Displayed Columns list.
6. Click Exit to return to the Set Query dialogue box. If you want to
save your query, see Saving a Query on page 82 of the Working
with SOFTIMAGE|3D User’s Guide for more information.
7. Click Ok to run your query. The Spreadsheet displays a column

SOFTIMAGE|3D Rendering 13
Introduction

for editing the projection plane values, and enabling and

disabling the cmpact (camera projection plane active) parameter.
Middle-click on a cell to toggle this parameter on and off.

Camera Projection Planes

(Asymmetrical Viewing Frustum)
To match real-world camera viewing with the camera in
SOFTIMAGE|3D, the metaphor of a projection plane located some
distance from the focal centre of the camera is used. By changing a
camera’s projection plane parameters, you can create an
asymmetrical camera in SOFTIMAGE|3D. These projection plane
parameters (such as distance, size, and offset) are available only in
the Spreadsheet window.
The following information supplements the information found in
the Animating User’s Guide and the Reference Guide.
A “normal” computer generated camera’s viewing frustum is
symmetrical; that is, if the field of view is n degrees, the top of the
frustum is angled n/2 degrees up and the bottom is n/2 degrees down
from the viewing direction. The horizontal angles of the frustum also
share symmetry and can be determined based on the vertical field of
view and the aspect ratio. An asymmetrical camera does not have
these restrictions.
You can specify an asymmetrical camera in SOFTIMAGE|3D using a
“projection plane” representation. That is, you can choose to specify
the distance to an imaginary projection plane, as well as its size in X
and Y and its offset from the viewing direction in X and Y. Note that
even when you offset it, the imaginary view plane always remains
perpendicular to your viewing direction.
Here are some ways in which you would use an asymmetrical camera.

Example 1
You are trying to model a real camera where the film (or the CCD, if
it’s a video camera) is not perfectly centred behind the lens. This is
often the case when trying to match computer-generated images
with real-world images.
Taking the example of a 35mm camera with a 50mm lens, the film
(projection plane) size is 24 x 36mm (its aspect ratio is 1.5).
However, suppose the film in your camera was offset by 1mm
horizontally and -0.5mm vertically from the true centre of projection

14 Rendering SOFTIMAGE|3D
 The Camera

of the lens. How would you model this accurately in

SOFTIMAGE|3D? You first open the Spreadsheet window and do the
appropriate query to get the Camera Projection Plane parameters
(see page 86 in the Reference Guide for how to do this). Then set the
following camera attributes:
• cmpdst (projection plane distance): 50
• cmpszx (projection plane x size): 36
• cmpszy (projection plane y size): 24
• cmpofx (projection plane x offset): 1
• cmpofy (projection plane y offset): -0.5
• cmpact (proj. plane representation active): TRUE
Also be sure to set the aspect ratio (cmaspr) to 1.5 to match your
projection plane’s ratio.
Note that you can also do strange 2D squash and stretch effects by
changing the projection plane’s X and Y sizes relative to one another
while keeping the viewport’s aspect ratio fixed at some value.
Offsetting the projection plane’s position in this manner is not
equivalent to slightly rotating the camera to compensate: the view
direction does not change, only the frustum’s shape does. If you
choose the Camera ¹ Show Camera and Show ¹ Cone (Unselected)
commands, you can see how the camera’s viewing frustum changes
when you change the values.
Another way to visualize an asymmetrical projection is to think of it
as a subregion rendering of a larger image. Changing the X and Y
offsets is equivalent to doing a 2D pan on the image (or moving the
subregion around), and changing the X and Y size is equivalent to
doing a 2D zoom on the image.

Example 2
Asymmetrical viewing frustums are also necessary to do correct
stereo perspective projections. The typical “improvised” method of
rotating the camera a bit to the right for the left eye and a bit to the
left for the right eye provides an adequate approximation, but
producing correct stereo projection images requires the use of
asymmetrical frustums.

SOFTIMAGE|3D Rendering 15
Introduction

16 Rendering SOFTIMAGE|3D
 C H A P T E R T W O

Rendering Basics

SOFTIMAGE|3D Rendering 17
Rendering Basics

18 Rendering SOFTIMAGE|3D
 Previewing before Rendering

Previewing before Rendering

Before you render a scene, you will probably want to preview it first.
Previewing lets you render a scene but not save it to file. It uses a
modified version of general rendering settings so that your
previewed scene is a fairly good approximation of the final look.
If you have not assigned material attributes to the object, it is
rendered with its default parameters. If you have not defined a light
source, SOFTIMAGE|3D creates a default light.

Setting Up for Previewing

Before you preview a scene, you should choose the Preview > Setup
command in the Matter module and enter an image resolution and
select the Preview Renderer (either SOFTIMAGE or mental ray) at
the minimum. The attributes you set up here also affect the previews
you do from the Material, 2D Texture, and 3D Texture dialog boxes.
Note: The mental ray renderer is not a part of the standard
Note
SOFTIMAGE|3D GT package, but may be purchased
separately.

Previewing in Different Ways

You can preview different objects or areas in a scene, or the whole scene.
• The Preview > All command in the Matter module allows you to
preview the whole scene.
• The Preview > Selection command module allows you to preview
only the selected object (or objects) in the scene. This is obviously
much quicker than previewing a whole scene if you don’t need to.

Previewing a Subregion
To preview a subregion, you can define an area in a scene according
to the parameter settings defined using the Preview > Setup
command. Objects in the defined area are rendered with material
attributes, textures, and lights. Hidden objects are not rendered.
The subregion cropping rectangle is displayed in window B, which is
the default Perspective window. If you have another window selected
(such as Schematic, Right, Front, etc.) when you choose this
command, SOFTIMAGE|3D switches it to the Perspective window.

SOFTIMAGE|3D Rendering 19
Rendering Basics

To preview a subregion:
1. Choose Preview > Setup. In the Subregion area, select the Draw
Subregion on Previewing option.
2. Click the Draw Subregion button. A red box appears in window
B that defines the region you want rendered.
3. To size the cropping rectangle, click on the sides or corners of the
rectangle and drag the mouse. When satisfied with the size,
release the mouse button.
4. To move the rectangle, click inside it and drag. When satisfied
with the position, release the mouse button.
5. The Preview Setup dialog box reappears. Click Exit.
6. Choose the Preview > Subregion command so it is active.
7. Choose Preview > All or Selection and then middle-click.
The preview appears on the screen showing only the subregion
you selected.
You can also define the subregion by assigning values to Subregion
Pixel and Percent (left, right, bottom, top).
For more information on subregion rendering, see Rendering a
Subregion on page 36.

Previewing a Hidden Line Image

In addition to previewing a fully shaded image, you can preview a
Wireframe
hidden-line picture in any window before you render it, which is
unlike the Line > Show command in the Tools module which
requires a rendered .lin file.
The command you choose (Faceted or Smoothed) corresponds to
the type of hidden-line renderer you can select in the Render Setup
dialog box.
Hidden-line
faceted Faceted means that lines are drawn for all the object’s surface
polygon edges. This makes the object look like a typical wireframe
object, except that the hidden lines are removed.
Smoothed means that the hidden lines are also removed, but only
the lines at the object’s edges are drawn, which creates an outline of
Hidden-line the object with no lines inside it.
smoothed

20 Rendering SOFTIMAGE|3D
 Previewing before Rendering

To adjust the way the hidden-line is processed for polygon mesh objects,
select the Automatic Discontinuity option in the dialog box that
appears when you choose the Info ➔ Selection command. The
Edge_Flag menu commands in the Matter module also let you adjust the
hidden-line process (see Showing and Hiding Object Edges on page 39).
To preview a hidden-line image, follow these steps:
1. Select any window by clicking the letter identifier box (A, B, C, or
D) in the upper-left corner of the desired window. The window is
highlighted in red. If you don’t select a window, window B is
used.
2. Choose the Hidden_Line command in the Tools module
(Faceted or Smoothed relates to the type of hidden-line renderer
you can choose in the Render Setup dialog box).
The image is rendered and displayed in the window you selected.
3. Middle-click to close the window.

SOFTIMAGE|3D Rendering 21
Rendering Basics

Optimizing Rendering Time

Before rendering, it is important to think of ways to optimize
rendering time, since this is one of the most time-consuming parts of
creating your animation. By keeping efficiency in mind when
modelling objects, defining their material attributes, and creating
light sources, projects which might otherwise be very time-
consuming to render will become much more feasible.

Modelling
When modelling, try to minimize the total number of triangles in a
scene. You can do this starting from the earliest stages of the working
process. For example, if you are drawing a curve to be used for extrusion
or revolution, you should use the fewest points possible to create the
desired contour: the resulting object will have fewer triangles.
If you plan to include lots of smooth rounded objects in your scene,
you can model them as polygonal mesh objects, then simulate their
surface smoothness by setting the Automatic-Discontinuity option
in the dialog box opened by the Info > Selection command: if the
objects were extruded from triangles instead of circles, they would
have fewer triangles.
If you are working with patch or NURBS surface objects, you can
minimize the number of triangles for objects that are located further
away from the camera by lowering their geometric resolution, again
by using the Info ¹ Selection command and its dialog box for patch
models. Also, instead of modelling very complex objects that have
many triangles, you can use texture mapping techniques for
transparency and roughness to simulate the desired effects.

Material Attributes
You can further optimize rendering time by keeping material
attributes such as reflectivity, transparency, and refraction, as well as
raytracing depth, and the number of lights and shadows in a scene to
a minimum. When used in combination, these attributes can
increase rendering time significantly. To save rendering time you can,
for example, apply a reflection map without raytracing to simulate
reflectivity. You can also apply reflectivity or transparency to only
one or two objects in a scene rather than to all of them. If you need to
increase raytracing depth, perform tests to determine the actual
required depth. Also, remember that each light you add to a scene

22 Rendering SOFTIMAGE|3D
 Optimizing Rendering Time

also affects the rendering: in general, most lighting effects can be

created using two or three lights, with only one casting a shadow.

Using mental ray

mental ray is a high-quality, photo-realistic renderer available in
SOFTIMAGE|3D. This renderer allows you to perform many special
effects as part of the rendering process instead of having to create
these effects in the scene itself. This helps to keep your scenes as small
as possible. This feature of mental ray also lets you save the settings
you used during one rendering process and use them in another one.
For more information, see Rendering with mental ray Software on
page 65 of the Reference Guide.

Compositing
Compositing is another very practical way to optimize rendering
time. By rendering constant elements of a scene, such as the
background, separately from other elements then compositing them
together afterwards, you can save a lot of rendering time. The time
saved is even greater if raytracing is required. The number of layers
that can be composited is unlimited. For more information, see
Compositing Images on page 111 of the Using Tools User’s Guide.

Render/Memory Guidelines
These are general rendering guidelines and should not be taken as
system limitations because there are ways of maximizing memory
usage or adding to available memory.
SOFTIMAGE|3D uses a Binary Space Partitioning (BSP) scheme (for
raytracing) which needs to “see” an entire scene to compute the rays’
trajectories. This allows for fast rendering, but very large scenes can
put large demands on memory.
Memory can be increased by adding RAM (Random Access Memory)
or by partitioning the disk allotting more disk space for swap space
(virtual memory). Because of the relatively slow access time of reading
information from disk, a scene that requires disk swapping renders
more slowly than a scene that fits into RAM. Swapping to disk varies in
degrees of relative slowness, depending on where and how often the
system needs to retrieve information from the disk.

SOFTIMAGE|3D Rendering 23
Rendering Basics

General Memory Requirements for Rendering

A general rule for calculating memory requirements is to allow 1
Mbyte of memory for every 1,000 surface triangles. For every 512 x
512 texture, you would need an additional 1 Mbyte of memory.
These rules assume that you’re using raytracing features (reflection,
shadows, refraction). If you are not using raytracing, you can render
much larger scenes.

Fitting a Scene into Memory

If a scene won’t fit into memory, there are a number of ways to
correct the situation. The simplest way is to increase the available
memory of the computer (RAM and/or virtual).
If increasing the amount of available memory is not possible, there
are ways of reducing memory demands. One easy way to reduce
memory demands is to use scanline rendering, meaning to not use
features which require raytracing.
Another common way to reduce memory demand is to break up
large scenes into separate, smaller scenes that can be rendered
separately. The rendered images can then be composited back
together. Memory demands can also be reduced by adjusting the BSP
(Binary Space Partitioning) tree which may, however, affect
rendering speed adversely.
Textures are a good place to look to for reducing memory
requirements. Be sure to use images that are as small as possible, yet
still give an acceptable result. No matter how much disk space images
take up, they all use the same amount of RAM depending on the
pixel resolution of the image. A 400 x 400 pixel image would use
400 x 400 x 4 = 640,000 bytes or 640 Kbytes of RAM.
A good trick for saving memory with texture and reflection maps is
when the colours are the same across the image, you could composite
the image so that it is only one pixel wide (this saves a lot of
memory). SOFTIMAGE|3D automatically “pulls” the texture across
the length of the texture-mapped surface. Texture maps can be quite
small with very acceptable results by using the Pixel Interpolation
option. This allows the renderer to average the in-between pixels to
give the effect of a higher definition map.
Tip:
Tip The Edit button in the 2D Texture File dialog box displays
the Cropping Utilities dialog box in which you can crop and
save a new picture file directly to the disk again.

24 Rendering SOFTIMAGE|3D
 Rendering a Sequence

Rendering a Sequence
To render an object or scene and save the image to file, you can use
the Render menu command in the Matter module or the standalone
renderer that is run from the command line in a shell.
With the Render menu command, its accompanying Render Setup
dialog box allows you to specify the type of renderer to use, the file
output name, the frame numbers to be rendered, image resolution,
antialiasing filter, pixel ratio, motion blur, and raytracing depth, as
well as other options.

Choosing a Renderer
To choose a render type, choose the Render menu command and
select a type from the Rendering Type menu. Here’s a brief
description of the various renderer types available to you:

Softimage Renderer
The SOFTIMAGE renderer is the default renderer. The extension of the
output file is .pic. The SOFTIMAGE renderer uses most of the parameters
in the Render Setup dialog box so that you can perform complete
rendering tasks. You can also run this renderer from the command line
(see Running the Renderer from the Command Line on page 31).

mental ray
The mental ray renderer is a high-quality, photorealistic renderer. It
provides an extensive set of built-in functions and can be
dynamically linked with user-defined shaders during the rendering
process. You don’t have to use shaders with mental ray, but there are
many different types of shaders which you can use to create
procedural textures (including bump and displacement maps),
materials, camera lenses, atmospheres, light sources, etc.
You can choose either mental ray 1.9 or mental ray 2.1. Note that
mental ray 2.1 contains features that mental ray 1.9 does not, such as
caustics and global illumination.
After selecting mental ray from the Rendering Type list, you can set
the parameters specific to it. These are accessed by selecting the
Antialiasing and Motion Blur options, as well as the Options
button.

SOFTIMAGE|3D Rendering 25
Rendering Basics

The extensions of the output files available for mental ray are:
.qntntsc (Abekas NTSC), .qntpal (Abekas PAL), .ct16 (RGBA 16
bits), and .rgb.
For more information on rendering in mental ray, see Rendering with
mental ray Software on page 65.

Hidden Line (Faceted or Smooth).

The hidden line renderer removes the lines that are usually hidden
from view in a wireframe object. This results in a more realistic
looking object than if you used the wireframe renderer (which shows
all object lines). The image is saved in .lin file format.
• If you select the Faceted option, the renderer draws lines for all the
object’s surface polygon edges. This makes the object look like a
typical wireframe object, except that the hidden lines are removed.
• The Smoothed option also removes hidden lines, but draws lines
only at the object’s edges, which creates an outline of the object
with no interior lines.

Wire Frame
The wireframe renderer renders wireframe images, which are made
up of the edges of objects and drawn as lines (this is the default view
shown in the windows when you open SOFTIMAGE|3D). This
renderer displays tracing features, such as edges or contour lines
without attempting to remove invisible or hidden parts, or fill
surfaces. The resulting image is in .lin file format.

Depthcue
The Depthcue renderer provides colour wireframe rendering. The
portion of the model closest to the camera appears “brighter” in
colour, giving the sense of depth in the z-axis. It is usually used to
derive more visual feedback from the models displayed on a screen
because the hardware shading is often too slow for complex models
and scenes.
Since the alpha channel is not calculated in the depthcue renderer,
compositing must be done using an external compositor, such as
SOFTIMAGE®|Eddie.

26 Rendering SOFTIMAGE|3D
 Rendering a Sequence

Hardware Renderer
The hardware renderer renders the same view you see when you
choose the Shade view mode in a window. The resulting image is
saved in .pic file format. The image produced is of lower quality;
objects show colour and lighting effects, but not shadows, reflection,
or transparency.

SOFTIMAGE|3D Rendering 27
Rendering Basics

Note: Remember to disable your screen saver while rendering with

Note this renderer. This process takes a snapshot of the screen
when the image is rendered.

Ghost
The Ghost renderer renders the same view you see when you choose
the Ghost view mode in a window. It allows you to display a series of
frozen snapshots of your animated objects at the current frame, next
frames, and previous frames. These frames remain in the view
regardless of the position of the time line pointer and are used as
reference point.

Rotoscope (Wire and Shade)

The Rotoscope renderer renders the same view you see when you
choose the Rotoscope Wire or Shade view mode in a window.
Rotoscoping is a technique in which individual frames of a video or
film are imported into SOFTIMAGE|3D and “traced.”
The 2D background can be a static picture or a sequence of images.
The scene can be zoomed and panned while maintaining perfect
registration with the background.
When zooming in window B, you must use the rectangle zoom
option (Shift-z Supra key) to define the zoom area; this takes a
snapshot of the rectangular area and magnifies it to fill the screen.
The camera position does not change. You can then use the zoom
option (z Supra key) to pan, zoom in, and zoom out on the defined
area. When you pan, it has the effect of moving the snapshot in front
of a still camera. When you select one of the Rotoscope options, the
Rotoscopy dialog box is displayed, as well as the window type in
which you are displaying the scene.

28 Rendering SOFTIMAGE|3D
 Rendering a Sequence

Rendering a Scene
When rendering with the Depthcue, Hidden Line, Wireframe,
Hardware, Ghost, or Rotoscope (Wire or Shade) rendering types, you
can use any of the displayed windows. Rendering with the other
rendering types must be done using window B with the Perspective view.
To do a basic rendering of a scene, follow these steps:
1. Select the window you want to render by clicking the letter
identifier (A, B, C, or D) icon in the upper-left corner of the
window’s title bar. The selected window is outlined in red. If no
window is selected, rendering is done in window B.
2. Choose the Render menu command in the Matter module. The
Render Setup dialog box is displayed. The parameters described
in this chapter refer to this dialog box.
3. Select a Rendering type (one of the previously described renderer
types).
4. Specify the sequence by setting the Start and End frame, and the
incrementation step.
5. Leave the default resolution as it is, but you can change it as
described in the next section.
6. Specify the name of the scene that you want to render by clicking
the Select button in the Output Image area and selecting a file
from the browser that appears. You can also type a name by
which you want to save the rendered image.
7. If you have the choice of file formats, select one from the File
Format menu.
For a very basic rendering, this is all you really need to set. If you
want to set raytracing, antialiasing, motion blur, or other more
sophisticated options, see Advanced Rendering on page 43.
8. Click the Render Sequence button to start the rendering process.
By default, the rendered picture is saved to the RENDER_PICTURES
chapter of your working directory.
9. To view the rendered image or sequence, see the Using Tools
User’s Guide for information on different viewing commands and
utilities available.

SOFTIMAGE|3D Rendering 29
Rendering Basics

Setting the Resolution

The resolution sets the frame resolution for your image in pixels. A
higher resolution produces a more detailed image.
The higher the resolution, the longer it takes to render. There is no limit
for image resolution, but if a resolution is greater than the monitor
screen (x = 1280), it cannot be displayed on screen while rendering.
In the Render Setup dialog box, you can assign the x-resolution the
value of 720 (Specify X) for the NTSC television standard. Once you
specify the x-value, the y-value is automatically calculated according
to the x-resolution, so when x is 720, the y-resolution is 486.
You can also set the y-resolution independently of the x-resolution
by selecting the Specify Y option. This command is more appropriate
if, for example, you are printing the image on paper. If you are doing
an advertisement that has unusual dimensions, you can set the x and
y-resolution values separately.

Resuming an Incomplete Rendering Sequence

If, for some reason such as a power outage or other interruption, you
have to interrupt the rendering process, you can continue rendering
at a later time using the Incomplete Rendering Sequence option. You
can resume rendering an incomplete rendering sequence, rendering
unfinished frames, and frames not yet started.
This is how to resume an incomplete rendering sequence:
1. Click on the Options button in the Render Setup dialog box. This
opens the Options dialog box.
2. Select on the Complete Unfinished Frames option.

Adding a Time/Date Stamp

The Time/Date Stamp button in the Render Setup dialog box stamps
the hour, month, day, and year of the position of the Sun light type in
the scene as it is rendered. This feature is useful for sending images to
a client, or keeping track of projects. This option is available only
with the SOFTIMAGE renderer.
When you click the Time/Date Stamp button, the Time/Date dialog
box appears. Select Active and pick the corner and the colour of text
you want the stamp to be. When you render, this information shows
up where you specified.

30 Rendering SOFTIMAGE|3D
 Rendering a Sequence

Running the Renderer from the Command Line

When you type soft -R instead of soft when starting
SOFTIMAGE|3D, the standalone SOFTIMAGE renderer loads. This
allows you to render an image without loading all of
SOFTIMAGE|3D.
The following example shows a rendering of the scene
RENDER-test.1-0.dsc in the local database. The -j option
allows more than one computer to render the scene without having
two computers render the same frame. The -s option specifies which
frames in the sequence to render and the sequence step, bypassing
the original setup.
soft -R RENDER-Test.1-0 -d local -j -s 10 352 1

The following example renders the latest version of the

Render-Plan3 scene, which is in the final database. The rendered
frames are recorded in /EX8a/RENDER/Plan3 and will be named
back3. The rendering can be run on more than one computer
because of the -j option. The ampersand (&) sends the process to
the background so that you can use the current shell to work on
other tasks.
soft -R RENDER-Plan3 -d final -L -j
-o /EX8a/RENDER/Plan3 -n back3 &

SOFTIMAGE|3D Rendering 31
Rendering Basics

Using the Standalone Renderer

If you’re using the standalone renderer (soft -R), one of its main
advantages it that you can render in the background while you
continue working.
There are two standard ways to send an IRIX process to the
background. The first way is to pause the process by typing Ctrl-Z,
then move the paused task to the background by entering the bg
command. For example:
soft -R final_logo_job -d tutorials -L -s 1 450 1
^z
bg
Tip:
Tip The Ctrl-Z command can be useful to simply pause a
rendering in progress so that you can do something else,
such as perform IRIX file management, etc., without the
huge overhead of a heavy rendering job using up all of the
CPU cycles. To return the job to the foreground, enter the
fg command.

The second way to send a process to the background is to add an

ampersand (&) to the end of the command. For example:
soft -R final_logo_job -d tutorials -L -s 1 450 1&

These two techniques put the process into the background, and allow
the current shell to be used for other things. The process continues to
run until the window they are running in is terminated.
If you desire the process to continue after the window is terminated,
or even after logging out, use the nohup command. To do this, create
an alias such as:
alias doit nohup soft -R

To run your alias, use the following syntax:

doit filename >& logfile &

For example:
doit final_logo_job -d tutorials -L -s 450 1 >& log &

32 Rendering SOFTIMAGE|3D
 Using the Standalone Renderer

filename represents a scene and logfile represents the text file to

which messages are redirected. It is important to redirect messages to a
file because messages are normally sent to the local window. However,
if the local window does not exist because it has been terminated or if
you have logged out, this will cause the process to stop.

Command Line Options

[-p <pre-frame script path>]

Specifies the pre-frame script that is executed.

[-P <post-frame script path>]

Specifies the post-frame script that is executed.

-m [ <base-outfile-name> ] ]

Forces the rendering to a single output file.

The base name of the outfile is the one stored in the
Render Setup dialog box.
An optional parameter (OUTFILE) allows you to
specify the base name in the form filename[.mi].
The .mi extension is optional and will be added if it
does not appear in the specified file name.
-M [ <base-outfile-name> ] ]

Same as -m, but creates multiple .mi output files.

One output file is produced per frame.

Usage
soft -R my_scene.1-0.dsc -m <filename>

Renders a single output file such as filename.mi

soft -R my_scene.1-0.dsc -M my_outfile

Renders multiple output files such as:

filename.1.mi
filename.2.mi
... up to filename.N.mi assuming a render sequence of frames 1 to N.

SOFTIMAGE|3D Rendering 33
Rendering Basics

Writing Rendering Scripts

Rendering scripts allow you to use C-shell scripts to do extra
processes before, or after each frame. These are used with the Pre-
frame and Post-frame Script options in the Render Setup dialog box.
The advantages of writing rendering scripts are numerous because
they automate tasks. Here are some of the advantages:
• They can be used to save disk space by dumping files.
• They can be used for processing and conversion.
• Post-frame can convert the file format as it runs.
• Once a file is rendered, you can use a script to start a composite
using a standalone; for example, you can use them to start Painterly
Effects as a standalone and then modify the images as they are
rendered.
When writing rendering scripts, Pre-frame executes the specified
C-shell before rendering an image and Post-frame executes the
specified C-shell script after the rendering of an image. Note that
arguments are passed to the program: $1 is the frame file name (.pic
sequence name); $2 is the current frame number (.pic frame); $3 is
the frame counter of the sequence; and $4 is the field. A C-shell script
must start with !# /bin/csh -f to define it as a C-shell; otherwise, it is
assumed to be a Bourne shell.
Tip:
Tip A script file needs to have execution permission. Type
chmod 777 and the file name to give the file all the necessary
permissions.
The following is an example script of parameter passing.
If you have the following file:
echo $1 $2 $3 $4 $5
and you specify this file as a Pre-frame or Post-frame script in the
Render Setup dialog box, and then render in the database /obladi to
the picture name blada from frames 1 to 3 without field rendering,
you’ll get this output in the shell from which you started
SOFTIMAGE|3D:
/usr/people/you/obladi/RENDER_PICTURES/blada 101
/usr/people/you/obladi/RENDER_PICTURES/blada 2 1 1
/usr/people/you/obladi/RENDER_PICTURES/blada 3 2 1

34 Rendering SOFTIMAGE|3D
 Using the Standalone Renderer

If you render the same thing from frames 11 to 13, you’ll get:
/usr/people/you/obladi/RENDER_PICTURES/blada 11 0 1
/usr/people/you/obladi/RENDER_PICTURES/blada 12 1 1
/usr/people/you/obladi/RENDER_PICTURES/blada 13 2 1
If you render the same file with field rendering on, with the
dominant field even or odd, you’ll get:
/usr/people/you/obladi/RENDER_PICTURES/blada 11 0 1
/usr/people/you/obladi/RENDER_PICTURES/blada 11 0 2
/usr/people/you/obladi/RENDER_PICTURES/blada 12 1 1
/usr/people/you/obladi/RENDER_PICTURES/blada 12 0 2
/usr/people/you/obladi/RENDER_PICTURES/blada 13 2 1
/usr/people/you/obladi/RENDER_PICTURES/blada 13 0 2
If you specify any parameters after the name of the file in the Pre-
frame or Post-frame script text boxes, the script finds these at the
beginning, not the end, of the variable list provided to the script. For
example, if you specify the following path in the script text box:
/path/example_script hello
the example script gives:
hello/usr/people/you/bladi/RENDER_PICTURES/blada 101
hello/usr/people/you/bladi/RENDER_PICTURES/blada 2 1 1
hello/usr/people/you/bladi/RENDER_PICTURES/blada 3 2 1
If you specify the following path:
/path/example_script hello goodbye
the result is:
hello goodbye/usr/people/you/obladi/RENDER_PICTURES/blada 1 0
hello goodbye/usr/people/you/obladi/RENDER_PICTURES/blada 2 1
hello goodbye/usr/people/you/obladi/RENDER_PICTURES/blada 3 2
The doesn’t show the field number at the end of each line because
the example script only prints out the first five arguments.

SOFTIMAGE|3D Rendering 35
Rendering Basics

Rendering a Subregion
If you don’t want to render your whole scene, you can render a portion
of it, called a subregion. There are two ways to do this: by either
choosing certain options in the Render Setup dialog box (Render
menu command) or by using the standalone renderer (soft-R).

Using the Render Menu Command

If you want to render a subregion using the Render menu command
in the Matter module:
1. Choose the Render menu command in the Matter module.
2. In the Render Setup dialog box, select the Subregion option to
click the Set button. The subregion only needs to be defined
once, since the Preview and Render processes share the same
subregion definition.
3. In the Define Subregion dialog box, select the exact area to be
rendered by entering either the pixel coordinates or by typing the
coordinates as percentages.
For example, if you want to test some material parameters, but only
want to preview the area affected by change and don’t want to
redefine the subregion for each test, you can set the pixel or
percentage coordinates for this area.
Note: The pixel values are dependent on the resolution. For
Note
example, if you set the subregion area with pixel coordinates
and the resolution changes (becomes larger or smaller), the
area that is rendered still uses those pixel coordinates.
However, the percentage values are independent on the
resolution. If you set the subregion area with percentage
coordinates, the pixel values change as the resolution
changes (if you set the percentage values to 10 percent, you
always see 10 percent of the image being rendered).
4. You can also draw the subregion to be rendered with the
cropping rectangle by selecting the Draw Always option. When
you select this option, the cropping rectangle is displayed in
window B, the default Perspective window. If another view is
selected in window B (such as Right, Front, etc.) when you chose
the renderer, SOFTIMAGE|3D switches it to the Perspective
window.

36 Rendering SOFTIMAGE|3D
 Rendering a Subregion

5. Resize the red rectangle to indicate the subregion you want to

render by clicking on a corner or side and drag the mouse. You can
move the rectangle by clicking inside of it and dragging the mouse.
6. Then click the Render Sequence button to render.

Using the Standalone Renderer

To render subregions using the standalone SOFTIMAGE renderer,
use the following format:
soft -R <scenename> -n output\
-b <distance from left edge>
<distance from right edge>
<distance from bottom edge>
<distance from top edge>
The subregion values should be in the range [0, 1] and be a
percentage of image resolution. They must overlap by 1 percent
because of the antialiasing of edges. To do this, use Adaptive
supersampling; (see Antialiasing on page 49). Also, note that the
values saved in the SETUP_SOFT.sts file work differently.
The following example renders four quadrants (whatever the
resolution may be):
soft -R test.2-0 -n a -b 0.00 0.51 0.49 1.00
soft -R test.2-0 -n b -b 0.49 1.00 0.49 1.00
soft -R test.2-0 -n c -b 0.00 0.51 0.00 0.51
soft -R test.2-0 -n d -b 0.49 1.00 0.00 0.51
The following example renders nine quadrants:
soft -R test -L -n a -b 0.00 0.34 0.65 1.00
soft -R test -L -n b -b 0.32 0.67 0.65 1.00
soft -R test -L -n c -b 0.65 1.00 0.65 1.00
soft -R test -L -n d -b 0.00 0.34 0.32 0.67
soft -R test -L -n e -b 0.32 0.67 0.32 0.67
soft -R test -L -n f -b 0.65 1.00 0.32 0.67
soft -R test -L -n g -b 0.00 0.34 0.00 0.34
soft -R test -L -n h -b 0.32 0.67 0.00 0.34
soft -R test -L -n i -b 0.65 1.00 0.00 0.34

SOFTIMAGE|3D Rendering 37
Rendering Basics

Later, you can simply composite the rendered images by defining the
full resulting resolution. The composite standalone uses the
information in the “comments” area of the picture file. A sample
composite script exists for nine quadrants resulting in a 1270 x 714
image. Execute this in the directory where the picture files reside.
composite -S 1270 714 final -d -v -s 1 1 1 a b c d e f g h
i

38 Rendering SOFTIMAGE|3D
 Showing and Hiding Object Edges

Showing and Hiding Object Edges

Hiding Edges
The Edge_Flag > Hidden/Rect hidden commands in the Matter
module allow you to define one or more edges of a polygon mesh
object as hidden when using the Hidden Line renderer. The hidden
edges are hidden regardless of the position of the object.
1. Select a polygon mesh object.
2. Choose the Edge_Flag > Hidden/Rect hidden command.
3. Pick the edges individually or perform a rectangular selection
using the left mouse button.
The edges are highlighted in cyan. The middle mouse button
deselects the edge, and the right mouse button toggles.
4. Press Esc to end the mode.

Selecting Hidden Edges

The Edge_Flag > Select/Rect select commands in the Matter module
allow you to select the edges of a polygon mesh object that are hidden
or visible when rendering using the Hidden Line renderer.
Note: This definition of edges is independent of the hidden-line
Note
process – a hidden edge is hidden regardless of the position
of the object.
1. Select a polygon mesh object.
2. Choose the Edge_Flag > Select/Rect select command.
The edges are highlighted in purple. The middle mouse button
deselects the edge, and the right mouse button toggles.
3. Press Esc to end the mode.

SOFTIMAGE|3D Rendering 39
Rendering Basics

Showing Edges
The Edge_Flag > Visible/Rect visible commands in the Matter
module allow you to define one or more edges of a polygon mesh
object as visible when using the Hidden Line renderer. The visible
edges are visible regardless of the position of the object.
1. Select a polygon mesh object.
2. Choose the Edge_Flag > Visible/Rect visible command.
3. Pick the edges individually or perform a rectangular selection
using the left mouse button.
The edges are highlighted in yellow. The middle mouse button
deselects the edge, and the right mouse button toggles.
4. Press Esc to end the mode.
Tip:
Tip The Show > Edge Flags command also shows or hides
visible or hidden edges. For curves and surface objects, these
are defined automatically upon creation.

Changing an Edge’s Discontinuity

The Edge_Flag > Discontinuity/Rect discontinuity commands in the
Matter module allow you to change the discontinuity of one or more
edges of a polygon mesh object locally, independent of the automatic
discontinuity angle set in the Info Selection dialog box. This allows you
to show the angularity of edges in the final rendered object.
1. Select a polygon mesh object.
2. Choose the Edge_Flag > Discontinuity/Rect discontinuity
command.
3. Pick the edges individually or perform a rectangular select using
the left mouse button.
The edges are highlighted in blue. The middle mouse button
deselects the edge, and the right mouse button toggles.
4. Press Esc to end the mode.

40 Rendering SOFTIMAGE|3D
 Viewing Rendered Images

Viewing Rendered Images

The Picture menu command in the Tools module allows you to view
a rendered image or a sequence of rendered images, or allows you to
get information about the image or image sequence.
1. Choose the Picture menu command.
The View Pictures dialog box is displayed.
2. Select one of the five options on the right side of the dialog box.
Click the Use icons option if you prefer to see pictures listed by
icon rather than by file name in the scroll box.
3. Click Display to display the file that you selected. If you selected a
Show option, a window showing that image appears. Middle-
click to close that window.
4. Click EXIT to close the dialog box.

SOFTIMAGE|3D Rendering 41
Rendering Basics

42 Rendering SOFTIMAGE|3D
 C H A P T E R T H R E E

Advanced Rendering

SOFTIMAGE|3D Rendering 43
Advanced Rendering

44 Rendering SOFTIMAGE|3D
 Introduction

Introduction
In addition to the basic rendering process, there are some rendering
features available that are not necessary for doing a basic render,
however, they can produce some useful effects. With such rendering
effects as raytracing, antialiasing, field rendering, or motion blur you
can produce effects that make your animation look much more
realistic, and give it a polished edge.
• Raytracing lets you render a scene with more realistic and precise
details, giving you a high quality render. You need to use raytracing
to calculate reflections, refractions, and shadows.
• Antialiasing is a method of smoothing out and sharpening rough or
fuzzy edges of graphics to produce a more polished look. This is
done by a mathematical process that subsamples pixels.
• Field rendering is used to reduce the strobing effect on fast moving
objects for rendering to video.
• Motion blur defines the relative blur of a moving object, usually for
special effects and for fast moving objects with lateral motion.
You can also render the Tag and Z channels of an image to use tag
channel and depth information, which is useful when compositing
and image-processing.
Details for each of these options is described on the following pages.
Each of these can be accessed from the Render Setup dialog box.

SOFTIMAGE|3D Rendering 45
Advanced Rendering

Raytracing
Raytracing calculates the light rays that are reflected, refracted, and
obstructed by surfaces. When rendering with raytracing, you will
have more realistic and precise results. However, it takes much longer
to render with a higher raytracing value.
If you have a scene with reflection, refraction, transparency, or
shadows in it, it will take a fairly long time to render. The beauty of
raytracing is that you can render what you want: if you want to see
only the effects of the animation, or just the effects of reflection, you
can deselect all of the other features and show only these options.
This will take less time to render, yet you benefit from the realistic
results that rendering with raytracing offers.
If there is no reflection or refraction in your scene, raytracing occurs
at about the same speed as hardware rendering.
The following illustrates when you would want to use raytracing
using the SOFTIMAGE renderer.
1. Select an object.
2. Choose the Material menu command in the Matter module. In the
Material Editor, select a glass material and apply it to the sphere.
3. Increase the Reflection and Transparency values.
4. Choose the Render menu command. The Render Setup dialog
box is displayed.
5. Select SOFTIMAGE as the rendering type.
6. Click the Options button. In the Options dialog box, set the
RayTraced Depth to 3 or 4. This means that the light rays will
bounce around the scene this many times.
7. Specify the sequence by setting the Start and End frame, and the
incrementation step.
8. Click Render Sequence and look at the results of rendering
with raytracing.
Note: If you want to put restrictions on the rendering process,
Note
such as only rendering with the reflection turned on, choose
Preview > Setup. Make sure Global On is selected and
choose which other parameters you want to have selected,
such as 2D Textures, reflectivity, Shadows, etc.

46 Rendering SOFTIMAGE|3D
 Raytracing

Raytracing in SOFTIMAGE|3D
SOFTIMAGE|3D raytracing uses a modification of the Glassner/
Kaplann spatial subdivision algorithm. In preprocessing the database,
first the bounding box (containing every object in the scene) is found.
This box is then divided in half on the x-axis. Both of the boxes are
then divided in half on the y-axis. The resulting four boxes are each
divided in half on the z-axis. These eight boxes are divided in half on
the x-axis… ad infinitum. This subdivision is halted when either:
• A box contains less than the preset number of triangles (triangles
per leaf)
or
• The number of divisions is greater than the preset maximum
(maximum tree depth).
In the rendering, the ray is passed from box to box until it hits
something. At each box, the ray must be tested against all contained
triangles. This triangle intersection testing is relatively slow
compared to the calculation of the next appropriate box, so it is to
your advantage to have a small number of triangles in each box.
However, as the number of triangles grows smaller, the amount of
memory required to hold all the boxes increases, as does the time
taken at the preprocessing stage.

Triangles per Leaf

Triangles per leaf allows you to optimize the raytracing process by
limiting the number of triangles per leaf that are rendered. Object
surfaces are divided into triangles to simplify rendering.
The smallest practical number for the Triangles per Leaf option is
10. You may need to increase this if the program runs out of memory
(that is, you get an error message like xsplit: malloc failed).

Max Tree Depth

The Max Tree Depth option defines the limit on the number of
subdivisions. It is necessary to prevent infinite recursion in the
preprocessing stage. Think, for instance, of the top of a sphere where
40 triangles touch. If the Triangles per Leaf is set to 10, the program
keeps on subdividing space around this point, trying to find a box
that contains only 10 triangles and failing. The Max Tree Depth
makes the software stop subdividing even though there are more
than 10 triangles in the box.

SOFTIMAGE|3D Rendering 47
Advanced Rendering

Finding the proper Max Tree Depth is more of an art than a science.
Empirical studies suggest that this is just beyond the point where the
number of leaves begins to decrease. To find this point, set the Max
Tree Depth to some large number (such as 50), select the BSP Tree
statistics in the Render Setup dialog box, and run the program (it is
not necessary to render a complete picture; the program can be
stopped when the first scan line appears). Exit the program and
examine the “stats” file. There should be a relatively obvious point at
which the number of leaves begins to decrease. Set the Max Tree
Depth to this number plus 1.
This is not an exact method, but by experimenting with the number
of Triangles per Leaf and Max Tree Depth, it is quite possible to find
a combination that results in a faster rendering time.

Reflection Mapping with Raytracing

Objects that are both reflective (have a Reflective value of more that
0 in the Material dialog box) and have an image assigned as a
Reflection Map (in the 2D Texture dialog box) use reflection
mapping. This takes the 2D image and wraps it around an infinitely
large sphere.
A reflection map without raytracing ignores the environment in
which an object exists. The 3D environment is not reflected on the
reflective object, only the reflection map picture will be. A reflection
map with raytracing reflects the 3D environment with the reflection
map behind it. If your reflective object is in a closed environment (in
a room) and you use reflection map with raytracing, only the room is
reflected; the room would hide the reflection map behind it. If you
need to reflect both the enclosed environment and the reflection
map, you should render twice and composite both layers.
For more information on reflection mapping, see Creating a Reflection
Map on page 122 of the Defining Materials and Textures User’s Guide.

48 Rendering SOFTIMAGE|3D
 Antialiasing

Antialiasing
Aliasing usually occurs when there is limited pixel resolution.
Antialiasing is a method of smoothing out and sharpening rough or
jagged edges of images to produce a more polished look. It uses a
mathematical process that subsamples the pixel area.
The number of pixels in a scene depends on the screen resolution.
The greater the resolution, the greater the number of pixels. The
smaller the resolution, the smaller the number of pixels.
The more pixels there are, less aliasing occurs and the edge is
smoother. When there are not enough pixels, you need to use
antialiasing to make the lines look smoother.
Tip:
Tip Avoid adding antialiasing when you render a texture map
that fills up the screen because its edges aren’t visible
anyway. If the texture map itself is already antialiased, you
don’t need to add more antialiasing when you render.
To use antialiasing with the SOFTIMAGE renderer, follow these steps:
1. Choose the Render menu command. The Render Setup dialog
box is displayed.
2. Select the SOFTIMAGE Rendering type. To use antialiasing with
the mental ray renderer, see page 88.
3. Specify the sequence by setting the Start and End frame, and the
incrementation step.
4. Select Antialiasing. The Antialiasing dialog box appears where
you can select either the Bartlett or the Adaptive supersampling
option as antialiasing method.
To help you make a choice on the type of antialiasing you should use,
here’s an explanation of the differences between Adaptive
supersampling and the Bartlett filter.

Bartlett
Bartlett is a static oversampling algorithm based on a variable width
box filter. As you increase the filter level, more of the adjacent
subpixel samples are taken into account while filtering a pixel. The
expense of this algorithm is caused by all pixels being supersampled
at the same rate (sample level) regardless of content affecting that
pixel; that is, a pixel with no geometry covering it gets sampled at the

SOFTIMAGE|3D Rendering 49
Advanced Rendering

same rate as a pixel with lots of geometry. One side effect of this
algorithm is due to using the box filter. As the sample rate (and thus
the width of the box) increases, more subsamples are taken into
account for a pixel; at high sample rates, this will sometimes lead to
“soft” looking images, meaning there are no well-defined edges in
the resulting image. Based on empirical data and practical limits of
time, a sample level of 4 is about the maximum advisable value.

Adaptive
For Adaptive supersampling, the algorithm is intended to decrease
the oversampling rate for those pixels that do not require it. This is
based on the contrasted thresholds specified. First, subpixel samples
are shot at the corners of each pixel. The variation in contrast is
evaluated and, if required, the pixel is subdivided into four subpixels
by sampling the full-sized pixel five more times, once in the centre
and once in the middle of each edge of the pixel. From here, each of
the four subpixels has the algorithm applied recursively, treating each
of the four as if they were actually a full pixel. This process continues
until the recursion limit is reached.

Visible Differences
You can see the difference between images filtered with the Bartlett or
Adaptive algorithm. These differences are primarily from the number
of subpixel samples shared between adjacent full size (result) pixels.
In the Adaptive algorithm, only the subpixel samples along the edges of
the full pixel are shared between adjacent pixels (that is, the four corner
samples are shared among the six adjacent pixels). Here’s a picture:
A B C D E

F G H I J
1 2

K L M N O
3 4

P Q R S T

U V W X Y

50 Rendering SOFTIMAGE|3D
 Antialiasing

This matrix represents image pixels: 1, 2, 3, and 4 are the first level
subsamples made for filtering pixel M. Samples 1 and 3 are also used
in the filtering of pixels G, H, L, Q, and R. Thus only these
subsamples are shared (along with any samples made between 1 and
3, are used for filtering pixel L, etc.).
In contrast, the box filter used in Bartlett can be viewed as a window
that moves along the scan line being filtered. All subsamples that fall
in the window are used to obtain the resulting pixel value for a given
pixel on the scanline. Assume we are filtering area M again with a box
filter that covers only two adjacent pixels: the letters now represent
subpixel samples, not image pixels. The window will cover all of the
subpixel samples A-Y, with M weighted most since it is central. Now
move the window to the next sample to filter N: now samples A, F, K,
P, and U are unused, but all the rest are used, as well as the samples
that are outside the diagram past E, J, O, T, and Y. So you can see that
subsample M still contributes to filtered pixels at both N and O. The
subpixel samples have an effect on filtered pixels that are half the
width away of the box filter.

Choosing an Algorithm
Results are usually a matter of preference.
The filtering effect of the Adaptive algorithm is slightly more
localized than the Bartlett. The reduction of computing time is often
found favorable and the filtering adequate with Adaptive. It only
recursively subdivides at the sub-pixel level, so the image will remain
crisper even at high filter levels.
If you don’t mind slightly “softer” looking images, and can deal with the
overhead of a static supersample, you will like Bartlett. Bartlett gives you
a progressively “softer” image the higher the filter level because it is
similar to blurring the image (more pixels are averaged into one).

SOFTIMAGE|3D Rendering 51
Advanced Rendering

Blurring a Moving Object

Motion blur allows you to define the relative blur of a moving object,
usually for special effects and for fast moving objects with lateral motion.
Note: Motion blur is used with animation rendered for film
Note
output to reduce the strobing. effect.

You can use motion blur from the Render Setup dialog box or from
the command line using the mb standalone program.
The following illustrates how to apply motion blur to an object using
the Render Setup options:
1. Select an object or objects.
2. Choose the Render menu command. The Render Setup dialog
box is displayed.
3. Select the SOFTIMAGE rendering type.
4. Select Motion Blur. The Motion Blur dialog box appears.
5. Enter the Shutter Speed which allows you to set the length of
time the shutter is open. Values are measured in frames.
6. Enter the Min. Movement which sets the sampling rate while the
shutter is open and the minimum amount of movement to be
blurred. The value is measured in pixels. A low value increases
the sampling rate and the memory required. The default value is
5.
7. Click Accept to return to the Render Setup dialog box.
You can also use the motion blur with the mental ray renderer. This
may be especially useful if you want to apply a motion blur to a single
object in your scene so it is much quicker. For more information on
using mental ray, see Blurring a Moving Object Using mental ray on
page 91.

52 Rendering SOFTIMAGE|3D
 Blurring a Moving Object

Using the mb Standalone

You can also create motion blur for objects in a scene by using the
mb standalone.
The following example applies a backward motion blur of 80% on
the test files, included in the final-roger.3.0 scene which is in the
ROGER database. The processed images (numbered from 1 to 98)
will be named testmb.
mb /usr/softimage3.5/rsrc final-roger3.1-0 -b -c .8 -d
ROGER -n test -N testmb -s 1 98 1
This example applies a backward motion blur of 100% on the saloon
images, numbered from 325 to 1082. Those images come from the
last version of the scene test-saloon, which is in the GEORGETTE
database. The processed images will be saved in /EX8a/RENDER/
saloon, and will be named saloonmb.
mb /usr/softimage3.5/rsrc test-saloon -b -L -d
GEORGETTE -n saloon -N saloonmb
-p /EX8a/RENDER/saloon -s 325 1082 1

For more information on the command line syntax for this

standalone, see the mb description in the Standalones HTML file on
the Online Library CD.

How Motion Blur Is Calculated

SOFTIMAGE|3D makes a copy of each object being blurred, and
each copy is updated at a different position in time in the animation.
As the number of requests for time steps increases, more object
copies are stored into memory which places high memory demands
on the system.
Note: Since most people use RAM to capacity, motion blur usually
Note
uses memory swap space for its calculations.

Tips for Using Motion Blur

Half Blur
You can choose to activate or deactivate the blur effect in front of or
behind an object. To produce a more traditional animation, you can
control the blur motion by activating its effect only behind the
moving object. Half Blur is available in only the mb standalone.

SOFTIMAGE|3D Rendering 53
Advanced Rendering

Alpha Channel
By default, resulting images will also have a blurred alpha channel. If
the sequence of animation is not to be composited, it is possible to
turn off the blurring of the alpha channel to reduce the time required
to finish blurring the sequence.

Depth Computation
To reduce computation time, you can disable use of the depth
information in the blurring process. This is only recommended when
the objects in motion do not cross paths.
• Motion blur is extremely handy because the colour element can be
rendered first. Make sure to render colour elements intended for
motion blurring using only small amounts of antialiasing, or no
antialiasing at all. This is because motion blur masks the aliased
edges in the resulting image anyway.
• Motion blur used on 3D procedurally texture-mapped objects
produces poor results.
• Motion blur has no effect on the apparent motion caused by
camera movements. This is due to the number of moving objects
that would over-extend the limits of the buffer memory.
Tip:
Tip If you use field rendering, you can produce a smoother
effect because you are doubling the number of frames by 2.

Memory Usage
Calculating motion blur can be very memory intensive. When the
message “malloc failed, reduce memory demands” is displayed, it
indicates that animated objects have exhausted the memory of the
system and that all available RAM and swap space have been exceeded.
This situation can be solved by using the following techniques:
• Limit rendering to one object at a time, and once completed,
composite the rendered objects together. Applying motion blur to a
single object usually achieves the desired effect.
• Reduce the physical size of the objects being blurred because the
number of replicated objects is a combination of object size, rendered
resolution, shutter speed, and the minimum movement values.

54 Rendering SOFTIMAGE|3D
 Blurring a Moving Object

• Render the objects requiring motion blur at half the original

resolution. Enlarge the objects with the zoom standalone to return
objects to their original size, and then composite them back over
the rest of the scene.
• Reduce or eliminate shadows, reflection, refractions, and
transparency in scenes because they use large amounts of memory
and reduce rendering speed.
Note: Usually, when the error message “malloc failed, reduce
Note
memory demands” appears, the only way to clear memory is
to exit the software and retrieve the scene; otherwise, results
that reflect the modified parameters may not take effect.
Even if physical memory limits are not reached, certain scenes may
still lack memory to complete rendering tasks. This is because of the
method the system uses to calculate its memory allocation before a
task is executed. The system calculates an estimated value of required
memory, and does not proceed if that estimate exceeds the system’s
current memory capacity.

Limitations with Motion Blur

Extreme Rotation or High Curvature Movement
Objects that show extreme rotation or movement over a small
number of frames will show some perceptible artifacts. For example,
one rapidly rotating cylinder end placed at a perpendicular angle to
the camera will produce unrealistic motion blur. The true motion
would show the moving end of the cylinder sweeping a smooth arc
between the three position (X-1, X, X+1). This algorithm, however,
will draw a straight lines between the three positions, giving a more
angular appearance to the resulting blur.

First and Last Animation Frame

Since the motion blur algorithm bases its evaluation on the previous
and the next frames, a conflict arises since no previous frame exists
before the first frame and no next frame exists after the last frame in
an animation sequence. A solution is to add an extra frames to the
fcurve at the beginning and one at the end. For example, for an
animation that initially runs from frame 1 to 100, you would extend
the fcurve to include frame 0 and 101. This will provide the
algorithm the necessary position data to evaluate the blur motion to
include frame 1 and 100 of the animation.

SOFTIMAGE|3D Rendering 55
Advanced Rendering

Rapid Object Growth and Scale

Motion blur’s ability to interpret the one-to-one pixel correlation is
lost if an object grows rapidly relative to its frame size or if an object is
scaled up over a small number of frames. This loss is because that the
geometry covers fewer pixels at frame X than at frame X+1. Without
the pixel correlation information, the algorithm produces a visible
artifact that appears as “tears” along the edge of the blurred object.

Inaccessible Data
Since motion blur uses image-based information, inaccuracies can
occur from loss of data as objects enter and exit the frame. This is
because the object position data that originates outside the frame is
inaccessible to the motion blur’s algorithm.

Transparent Objects
Transparent objects may cause unexpected artifacts to appear in
blurred images. Scenes that contain moving transparent objects
passing in front of static objects create visible artifacts. The pixels
associated with the moving object are blurred even though the
colour contribution of those pixels is primarily from the background
object. Conversely, if a moving object passes behind a static
transparent object, the portion of the object visible through the
transparency will be blurred incorrectly.

Antialiasing/Inaccurate Colour Choices

An artifact will result from antialiasing that is used to create the
original in blurred images. When the colour of a pixel is derived
mainly from the background colour with a small colour contribution
from a moving piece of geometry, the blur algorithm incorrectly
includes the background colour as part of the moving geometry pixels.

56 Rendering SOFTIMAGE|3D
 Field Rendering

Field Rendering
Field Rendering is used to reduce the strobing effect that results from
fast moving objects when rendering to video. If you are rendering to
film, refer to motion blur (see page 52).
SOFTIMAGE|3D uses a wide-pixel rendering technique. Internally,
the camera’s field of view is perceived as being half as high, but each
scan line is amplified to twice its thickness. The sampling of the
pixel, therefore, is estimated over a larger area than normal. The
camera first considers the even and then the odd lines. It is raised or
lowered to ensure that the two different images have the correct
visual orientation. The direction of compensation depends on
whether your dominant field is set to even or odd. The following
figure illustrates the use of the wide-pixel rendering technique.

1 1
2
3 3
4

The following example shows how the scene is viewed by the camera at
render time. The direction of compensation corresponds to your
dominant field setting (even or odd). Each line is scanned in at 50%
wider than the normal sample line to ensure correct picture proportions.
regular field of view
field render sample pixel

*Note that the topmost

pixel is out of the regular
1/2 field of view in this
the number of illustration.
scan lines

1 pixel

There are two versions of the post-frame script file: inter01 and
inter02. File version inter01 corresponds to a dominant field setting
of 2,1 and inter02 is for a 1,2 setting. Select and use the post-frame
file version that corresponds to your dominant field setting. These
files ensure that merging occurs after each full frame (two fields) has
been processed. The post-frame script files invoke the standalone
program called interleave. The script files are located in
/usr/softimage/3D/tools/scripts/inter01.

SOFTIMAGE|3D Rendering 57
Advanced Rendering

Note: The inter03 and inter04 scripts perform the same function
Note as inter01 and inter02 respectively, except that they remove
the separate image fields after the composite frames have
been created.
To use field rendering with the SOFTIMAGE or mental ray renderer,
follow these steps:
1. Choose the Render menu command in the Matter module.
2. In the Render Setup dialog box, select either SOFTIMAGE
renderer or mental ray as the rendering type.
3. Select the Field Rendering option. The Field dialog box is
displayed.
4. Select the Active option, then select the Even or Odd option to
correspond to your Dominant field environment.
5. Click the Accept button.
6. Enter a Post-frame script in its text box, or click the Post-frame
button to open a browser in which you can locate the /usr
/softimage/3D/tools/scripts/inter01 file which is used to
combine the rendered frames. Select the post-frame script file
inter01 or inter02 that corresponds to your video editing
environment. You may want to copy this file to your working
directory.
Note: If the fields fail to combine after they have been rendered,
Note
verify to ensure that the appropriate ASCII file has been
entered in the Post-frame text box and not mistakenly
placed in the Pre-frame text box.

Pre and Post-Frame Scripts

These scripts allow you to use C-shell scripts to do extra processes
before or after each frame.
Arguments:
$1 = frame file name (.pic sequence name)
$2 = current frame number (.pic frame)
$3 = frame counter of the sequence
$4 = field

58 Rendering SOFTIMAGE|3D
 Field Rendering

Creating a Setup
These two steps will help you recognize the four possible resulting
field combinations produced by the post-frame script files:
1. Select the appropriate post-frame script file to suit your
dominant field setting.
2. Render a single frame from a merged sequence and record a few
seconds on tape. A cube moving over 10 frames with a stationary
patch sphere was used for this test. The cube is used to verify
correct field order, and the sphere is used to verify environment
type.
Compare your results with the following:
• A normal image where the correct field setting and appropriate
post-frame script were selected.
• If the cube appears to jump back and forth, the incorrect field setting
and inappropriate post-frame script were selected. To view this
clearly, use the VTR in slow frame-by-frame motion. Correct this by
changing your dominant field setting and using the alternate post-
frame script file to reverse the order of the frames when combined.
• If the cube appears broken at the top and bottom pixels, but the
motion is smooth, an incorrect dominant field setting and correct
post-frame script were selected. Correct this by changing your
dominant field setting.
• If the sphere is stable and the motion is not smooth, a correct
dominant field setting and incorrect post-frame script were
selected. Correct this by using the alternate post-frame script file.
Note: The Avanzar video output board has the ability to read and
Note
combine the fields as it records to tape. This may eliminate
the need to use post-script files, unless you need to evaluate
the quality of the combined frame or have access to the
flipbook display on the workstation.
Tip:
Tip It is recommended that NTSC users set the Dominant field
to ODD in the Render Setup dialog box, and save the Setup
File with the Preferences > Setup File > Save command. By
including the -s option at the end of the soft alias string, this
feature is automatically set to the correct position for each

SOFTIMAGE|3D Rendering 59
Advanced Rendering

new session. The Picture format and other preview and

rendering parameters will also be saved. Look at the ASCII
file to verify the various parameters available.

60 Rendering SOFTIMAGE|3D
 Rendering Tag and Z Channels

Rendering Tag and Z Channels

Tag and Z Channels provide additional channel information to
increase flexibility and enhance compositing and image-processing
results. You can use Z Channels with the composite standalone
program and Tag Channels with the Painterly Effects program for
post-processing.

Tag Channels
The information contained in the Tag Channels consist of 1-bit
stored layers that are identified by using the Select ➔ Set Named
Selection command specifying a name, and selecting the User Tag
option. The rendered file has the form of <name>.<frame number>
with a .tag extension instead of the usual .pic extension.

User Tag
Items to be included in a user tag selection. Tagged objects are
rendered into a separate file which contains a mask showing all pixels
in the image affected by the tagged model.
Note: You can define multiple objects within each Tag layer, but
Note
since there is no depth information, Painterly Effects
considers the whole layer as one section. Only one pixel is
designated as the “tagged” pixel. Assign objects that require
a specific effect to a separate layer. In a case where two
different scene images (layers) have the same tagged pixel,
the topmost layer will be used.

Using the Tag Channel Option

With the Tag Channel information, post-processing utilities (such as
Painterly Effects) can isolate pixels associated with a given object and
process only those pixels. For example, a post-processing effect such
as glow can be used to affect a specific part of an image or the entire
image. Without the tag information, the post-processing utility
would have to either locate the edges of a specific object or apply the
effect to the entire image.

SOFTIMAGE|3D Rendering 61
Advanced Rendering

The following example using three cylinders describes the typical

procedure to follow when using Tag Channels:
1. Create a “target” using three different-sized and different-
coloured cylinders.
2. Superimpose all three, placing the smallest one in front.
3. Select the second cylinder, then choose the Select ➔ Set Named
Selection command.
4. In the Set Named Selection dialog box, select the User Tag
option, then select the Create button.
5. Repeat the same operations for the third and largest cylinder.
Notice that you now have two user tag items in the Schematic
window.
6. Choose the Render menu command and render your picture
with the Render Tag Channel option selected.
7. Exit SOFTIMAGE|3D and open Painterly Effects.
8. Load your image and select an effect (such as Rough Pastels),
then click on the Tag Channels button.
9. Select Use Single Tag Channel and enter the number of the Tag
Channel to affect using the Rough Pastels effect (1 or 2).
10. Select Ok and preview.
You will notice that the effect affected only the object with the Tag
Channel 1. If you add a ripple effect on Tag Channel 2, it will only
affect the cylinder with Tag Channel 2.

Z Channel
The Z channel information allows for more advanced compositing
operations. Z channel provides depth information so that you can
position an object in front of and in back of the background image.
One useful application would be to use the Z channel information to
allow an object in a scene to interact with a background image.
Without the Z channel information, the compositer can only decide
which layer should be placed on top, confining selected objects to the
front of the background image.
Z channel information is also useful when compositing a
SOFTIMAGE|3D scene with particles created in the SOFTIMAGE®
Particle program.

62 Rendering SOFTIMAGE|3D
 Rendering Tag and Z Channels

Using the Z Channel Option

The following example of a computer-generated airplane flying
around a scanned image of a building describes the typical procedure
for using Z Channels:
1. Simulate the Z Channel for the background image using the
Rotoscopy view mode to create a simple model of the building.
2. Choose the Render menu command and render this model with
the Render Z Channel option selected.
3. Rename the .Zpic file created using the same base name as the
building image (such as stuff.Zpic or building.Zpic.) to inform
the compositer that the building image has an associated Z
Channel.
4. At this point, you need only ensure that the path of the airplane passes
behind the simulated position of the building, and then composite.
The resulting animation will show the airplane passing behind the
static image of the building.

Limitations
The quality of the Z compositing resolution diminishes when you
physically intersect any object. This is because the polygonal triangle
of one object is overlapped by the object placed before it. The
program averages and filters the colour pixel information, but
cannot do the same for object location in space. Consequently, the
resulting image will appear non-filtered at object intersection points.
Presently, it is impossible to antialias depth information due to the
program’s single sample per pixel calculation formula.

SOFTIMAGE|3D Rendering 63
Advanced Rendering

Rendering Faces of an Object

By default, only the front faces of an object are rendered; that is, the
ones whose normals are facing the camera. The back faces are “culled”
(ignored). If you like, you can render all faces of an object. This could
be useful if you are rendering something like a flag waving in the wind,
in which case you would want both sides of the flag to be rendered.
The faces that are rendered are defined by the Back Culling option in
the Render Setup dialog box.
1. Choose the Render menu command. The Render Setup dialog
box is displayed.
2. Select the SOFTIMAGE renderer as the Rendering type. To use
back culling with mental ray, see Rendering Faces of an Object on
page 106.
3. Click on the Options button.
4. Select Back Culling to render only the front face (the default), or
deselect it to render all faces.

64 Rendering SOFTIMAGE|3D
 C H A P T E R F O U R

Rendering with mental ray ® Software

SOFTIMAGE|3D Rendering 65
Rendering with mental ray Software

66 Rendering SOFTIMAGE|3D
 Introduction

Introduction
The mental ray renderer is a high-quality, photo-realistic renderer
available in SOFTIMAGE|3D. This renderer allows you to perform
many special effects as part of the rendering process instead of having
to create these effects in the scene itself, which keeps your scenes as
small as possible. This feature of mental ray also lets you save the
settings you used during one rendering process and use them in
another one.
mental ray gives you more versatility, better image quality, and the
ability to run on a network or on a multi-processor machine.
Another great aspect of mental ray lies in its open architecture which
gives you the option of writing and applying custom shaders.

Taking Advantage of a Network

The mental ray renderer allows several imaging possibilities which
were not previously attainable by the SOFTIMAGE renderer. One of
these is the capability to increase the speed of the rendering process.
Through network rendering, you can use other machines licensed
with SOFTIMAGE|3D to capture unused processor power. To do
this, simply create a file called .rayhosts in the home directory. This
file needs to contain the names of the other machines, each written
on a separate line. There should be a noticeably decreased time in the
rendering of each frame.
The simplicity of this method only works if every machine involved in
the network rendering is of the same type (all NT or all IRIX) and if
SOFTIMAGE|3D, shader libraries, texture libraries, etc, are all stored
in the same location on each machine. As they’re likely not, you must
use the linktab.ini file, which helps different machines with different
set ups talk to each other. For more information, see Distributed
mental ray Rendering between IRIX and Windows NT on page 117.
As each frame is rendered, tiles appear within it. These tiles are the
rendered portion of the scene sent back from the host machines. The
trade-off of parallel processing is that the scene information must be
sent to each of the host machines. This makes network rendering a
useful choice when the time necessary to render the frame is actually
greater than the time required to load and dispatch the scene data to
the rendering servers.

SOFTIMAGE|3D Rendering 67
Rendering with mental ray Software

Custom Shaders
The mental ray renderer provides an extensive set of built-in
functions and can be dynamically linked with user-defined shaders
during the rendering process. You don’t have to use shaders with
mental ray, but there are many different types of shaders you can use
to create procedural textures (including displacement maps),
materials, camera lenses, atmospheres, light sources, etc.
SOFTIMAGE|3D supplies sample shaders of each type, but you will
probably want to use mental ray to its full advantage and create your
own shaders (see the mental ray Programmer’s Reference Guide, which
is located on CD #3). You can use any combination of these shaders
to achieve specific results, such as using a lens flare lens shader for the
camera with a “star” shader for the light.

68 Rendering SOFTIMAGE|3D
 Introduction

Setting Up for Single and Multiple Objects

The mental ray renderer traces the entire scene like other rendering
software. However, instead of treating every object with the same ray
information, object “flags” can be set to ignore verbose rendering
information of specified objects. This means that you can use mental
ray’s options for only the objects that you choose, which can make
the rendering process go much quicker.

Render Options for Single Objects

To use mental ray on a single object:
1. Select a single object in the scene (this must be a 3D object such
as a polygon mesh or patch or NURBS surface object).
2. Choose the Info > Selection command. The Info dialog box for
the object type you selected appears.
3. Click the Render Setup button in this dialog box.
4. In the mental ray Render Setup dialog box that appears, set the
parameters for the object, such as motion blur, visibility, and surface
approximation (see Options for Selected Objects on page 70).

Render Options for Multiple Objects

To use mental ray for more than one object on a scene:
1. Select multiple objects with the Multi menu command active.
2. Choose the Info > mental ray command. This displays the
mental ray Propagation dialog box.
This box is identical to the mental ray Render Setup dialog box as
previously described except for two modes for editing the
parameters: the Edit fields values option lets you edit the state of the
selected parameters, but the Select/unselect propagated fields mode
is the default. This mode allows you to select only those parameters
that you want to apply to the selected models.
With the Select/unselect propagated fields mode selected, all the
options will be dimmed. However, you can select any options you want
to use by clicking on them (you can deselect them in the same way).

SOFTIMAGE|3D Rendering 69
Rendering with mental ray Software

The options in this dialog box are “smart” because they edit only the
parameters which relate to a particular type of object. If there is a
patch object and a polygon mesh object in the selected group, the
Propagation parameters edited will affect only the objects to which
they are applicable.
3. Select the Select/Unselect propagated field option for the Edit
mode. Pick the fields you want to edit and they become unghosted.
4. Change the parameters that you want (see the next section), and
click the Apply on Selected button. This applies the parameters
set to the selected objects, but leaves the dialog box open.
5. Click Exit to close the dialog box.

Options for Selected Objects

If the final rendering of the scene will use mental ray, these parameters
should be taken into account throughout the creative process.

Visibility of Objects
The three Object Visibility options determine if an object is visible
when rendered with mental ray. You can also choose to have a visible
shadow and reflection of that object. These options can be selected
individually or combined to achieve the desired effect. The default is
that all three options are selected.
• Primary rays refers to all rays emitted from the camera. It allows
you to specify the visibility of an object in the scene. When this
option is selected, the object is visible.
• Secondary rays allows you to turn an object’s reflection in a
mirrored surface on or off. When selected, a reflection is visible.
• Shadows allows you to turn an object’s shadow on or off. When
selected, a shadow is cast.
For example, if you create two characters (an angel and a devil) and
make them both walk past a mirror, you can set the Object Visibility
parameters so that only the angel is visible to the camera, but the
shadow and the reflection in the mirror belong to the devil. To do
this, select only the Primary rays option for the angel, and select only
the Secondary rays and Shadows options for the devil.

70 Rendering SOFTIMAGE|3D
 Introduction

For information on how to create shadow objects, see Creating Shadows

on page 34 of the Defining Materials and Texture’s User’s Guide.
For information on using the Secondary rays option for creating a
reflection map, see Creating a Reflection Map Using mental ray on
page 124 of the Defining Materials and Texture’s User’s Guide.

Motion Blur
The motion blur options allow you to specify varying degrees of
motion blur for a selected object. For information, see Blurring a
Moving Object Using mental ray on page 91.

Shadow Maps
You can enable mental ray shadow maps, which are fast
approximations of raytraced shadows. See Creating mental ray
Shadow Maps on page 99 for more information

Displacement Maps
You can also use mental ray to create displacement maps. See
Creating mental ray Displacement Maps on page 103 for more
information.

Surface Approximation
For information on Surface Approximation, see Surface
Approximation on page 108.

SOFTIMAGE|3D Rendering 71
Rendering with mental ray Software

Overview for Using mental ray

The mental ray renderer is used like any other renderer, but has a
many more options that can be set, which are not all found in one
place. This outline shows you where to find these new options and
describes how to use them to achieve the rendering results you want.
Tip:
Tip The first step in using mental ray is to inspect the scene to be
rendered. From the observations you make doing a triangle
count and checking material assignment, you can evaluate
how to best use the various options of mental ray.
1. Create the scene or object you want to render.
2. If you select a single object to render, choose the Info ➔ Selection
command. If you select multiple objects in the scene to render,
choose the Info > mental ray command (see page 69 for both these).
3. With the Light > Define command, you could set the Area lights
geometry options to bring more realistic effects to light sources.
The area lights model the light source as a piece of geometry,
which allows soft shadows to be created (see page 36 in the
Defining Materials and Textures User’s Guide).
4. You can also choose shaders for the different effects that you
want, such as camera lens shaders, light shaders, material shaders,
volume shaders, shadow shaders, and textures (2D and 3D)
shaders. You can use any combination of shaders to achieve
specific results, such as using a fish-eye lens shader for the camera
with a cloud volume shader for the atmosphere.
- Choose the Camera > Settings command to access lens shaders.
- Choose the Light > Define command to access light shaders.
- Choose the Material menu command to access material, volume,
and shadow shaders.
- Choose the Texture > 2D Global/2D Local command to access
2D texture shaders.
- Choose the Texture > 3D Global/3D Local command to access
3D texture shaders.
- Choose the Atmosphere > Depth-Fading command to access
volume shaders.
- Choose the Render menu command and select mental ray
Options to access output shaders.
5. Choose the Render menu command, select mental ray as the
rendering type in the Render Setup dialog box, and set all the
mental ray rendering options as described in this chapter.

72 Rendering SOFTIMAGE|3D
 Switching Between mental ray versions

Switching Between mental ray versions

SOFTIMAGE|3D includes both mental ray versions 2.1 and 1.9. You
can easily switch to one or the other from the Preview Setup and
Render Setup dialog boxes.

SOFTIMAGE|3D Rendering 73
Rendering with mental ray Software

Previewing with mental ray

If you want to preview your image in mental ray, you must change
the Preview Setup. This also sets up for previewing from within the
Material, 2D Texture, and 3D Texture dialog boxes, which is
necessary if you want to preview the effects of shaders from within
these dialog boxes.
1. Choose the Preview > Setup command in the Matter module.
2. Select mental ray as the Preview Renderer.
3. Click Ok.

mental ray Preview Settings

You can view a mental ray preview in a window. To select this option:
1. Choose Preview > Setup in the Matter module.
2. Select the mental ray renderer for previewing. A new Preview in a
Window option appears—select it.
3. Click the Preview button in dialog boxes that have it (many
dialog boxes now have this option, such as ToonAssistant).
4. Press Esc or middle-click in the preview window to return to the
dialog box.
When you activate this type of preview, the rendered picture is
displayed in a window. You can move this window at any time, even
while the picture is being rendered.
Note: When you have repositioned the preview window, the
Note
position remains the same the next time you preview an image.
It is not reset to the default positioning (centre of the screen).

74 Rendering SOFTIMAGE|3D
 Previewing with mental ray

Optimizing mental ray Previewing

The default settings for previewing optimizations are suitable for
most scenes, but there may be specific scenes for which the settings
are not optimal, or for which you would like to preview a single
object using the final rendering settings. The optimization option
allows you to override the default previewing optimization settings
in the same way that optimization settings are edited for rendering
1. Choose Preview > Setup in the Matter module.
2. Click the button beside Antialiasing. Both the SOFTIMAGE
renderer and mental ray can be optimized independently. The
following dialog box appears:

3. Select the Active option to activate the antialiasing parameters.

Select the options that you want to apply and click Accept. See
the SOFTIMAGE|3D Rendering User’s Guide for more
information on antialiasing.
• If the Active option is selected, the optimization parameters that
you have set in the dialog box are activated.

SOFTIMAGE|3D Rendering 75
Rendering with mental ray Software

• If the Active option is deselected, the default rendering settings are used.
Note: After you have specified the antialiasing settings, you can
Note
toggle the parameters on and off by middle-clicking the
Active option.

Speeding Up Rendering Time

You can greatly accelerate previewing by using mental ray’s
infrasampling capability. A rendered image is infrasampled when less
than one ray is cast for each pixel (antialiasing involves oversampling
the image; that is, casting more than one ray per pixel).
Infrasampling results in a lower quality image containing artifacts
(such as jagged edges), but it renders in less time. You can achieve a
quick rendering while sacrificing image quality by doing the
following:
• Setting the Min Samples parameter to a negative value (such as -2
or -3).
• Raising the Threshold values to high values (such as 0.5 or higher)
• Setting the Max Samples parameter to a relatively low value (such as 0).

76 Rendering SOFTIMAGE|3D
 Disabling Effects When Previewing and Rendering

Disabling Effects When Previewing and

Rendering
You can turn mental ray 2.1 features on and off, just like you can with
Softimage rendering effects. This will help speed up previewing and
rendering as you tune specific effects.

Choosing What to Preview

4. Choose Preview > Setup from the Matter module.
5. Select mental ray 2.x as the Preview Renderer.
6. Turn off any of the following effects:
- Area lights
- Lens effects
- Volume effects
- Output shaders
- Displacements maps
- Shadow maps

SOFTIMAGE|3D Rendering 77
Rendering with mental ray Software

About Previewing with the Shader Ball

Note that you can only use the Shaderball preview—which is
available in all mental ray shader dialog boxes—with shaders that
also have a mental ray 1.9 equivalent. This is because there is no
Shaderball preview available for mental ray 2.1. Instead, the 1.9
shader is used to preview the effect.
You can use the Shaderball with the mental ray shaders included with
SOFTIMAGE|3D Version 3.8 SP2.

Choosing What to Render

1. Choose Render from the Matter module.
2. Select mental ray 2.x as the Rendering Type.
3. Click the Options button and turn off any of the following effects:
- Area lights
- Lens effects
- Volume effects
- Output shaders
- Displacements maps
- Shadow maps
- Motion blurred shadow maps

78 Rendering SOFTIMAGE|3D
 Using Shaders

Using Shaders
Shaders are the key to adding special rendering effects to
SOFTIMAGE|3D. A shader is a simple program written outside of
SOFTIMAGE|3D and then accessed through the mental ray interface
in SOFTIMAGE|3D. A shader is an opening in the architecture of
mental ray which lets you program rendering variables. Instead of
choosing a pre-programmed shader, you could write your own
shader to achieve quality improvements, performance optimization,
or simply create a rendering effect not available in the shaders that
accompany SOFTIMAGE|3D.
Once compiled, the parameters of the programmed shader can be
edited as easily as editing the standard shaders. For more
information, see the mental ray Programmer’s Reference Guide.
SOFTIMAGE|3D ships with a variety of shader libraries. There are
several areas in its interface for applying a shader, depending on the
type of shader it is. There are shaders for material, volume, shadows,
2D and 3D textures, camera lenses, and lights, and atmosphere. The
output shaders perform post-processing on the rendered image.
• Choose the Camera > Settings command to access lens shaders.
• Choose the Light > Define command to access light shaders.
• Choose the Material menu command to access material, volume,
and shadow shaders.
• Choose the Texture > 2D Global/2D Local command to access 2D
texture shaders.
• Choose the Texture > 3D Global/3D Local command to access 3D
texture shaders.
• Choose the Atmosphere > Depth-Fading command to access
volume shaders.
• Choose the Render menu command and select mental ray Options
to access output shaders.
This is a general procedure for selecting a shader once you have chosen
the appropriate command. The exception for this is output shaders:
1. To select the shader to apply, Shader option in the mental ray
area of the dialog box you have open.
2. The browser opens to the chapter for that type of shader appears.

SOFTIMAGE|3D Rendering 79
Rendering with mental ray Software

In this example, it is the VOLUME_SHADERS chapter that is

automatically loaded.

3. Select one of the shaders and click the Load button. Its name
appears in the text box below the shader type’s name in the main
dialog box.
4. To edit the shader’s parameters, highlight the shader’s name in
the text box and click on the Edit button. The dialog box showing
all parameters for that shader appears.
5. If you want to save the shader by another name and then edit the
parameters, you can modify its name in the text box. This creates
a new shader with the new name, but with the current
parameters. You can then edit the new shader’s parameters as
described above.
The new shader can then be saved and recalled by name in other scenes.
Tip:
Tip Before previewing, go into the Preview Setup dialog box and
select mental ray as the rendering type.

80 Rendering SOFTIMAGE|3D
 Using Shaders

Linking a Custom Shader to a Script File

You can enter commands for the custom shader linking into the
mental ray file so that you don’t need to modify script files manually
to include link, code, and $include commands for non-standard
shaders. The syntax of the commands must be exact.
To link a custom shader, follow these steps:
1. Choose the Render menu command.The Render Setup dialog
box is displayed.
2. Select mental ray as the Rendering type.
3. Click the Options button. The mental ray Options dialog box appears.
4. Click the Edit button next to the Default Link/Code button to
display the Custom Script Line dialog box.
5. If you want to edit the command line, click the Edit button inside
this dialog box. The Custom Script Lines dialog box is displayed
again. Make any changes, and click Ok to exit the dialog box.
6. The Add button lets you add a new command to the script. Click
the Add button and the Append New Line dialog box appears.
You can add a maximum of 50 command lines.
7. Enter the new command line and click Ok to Accept and return
to the Custom Script Lines dialog box.
8. If you want to delete a command line from the script, click the
Delete button.

Lightning Volume Shader

You can specify multiple source and destination pairs for a single
instance of the lightning shader. This lets you can attach a model to a
single instance of the lightning shader, which in turn references any
number of source/destination pairs of lightning bolts. This makes it
possible to create complex lightning effects.
You cannot use the same model name more than once in either the
Source or Destination text box. If you want to create two bolts with a
common source point (null1), you would:
• Duplicate the source point (null1)
• Name it null1_dup

SOFTIMAGE|3D Rendering 81
Rendering with mental ray Software

This table illustrates some ways of specifying source/destination

pairs, with different results.

Source Destination Does this

null1 null2 null3 null4 Creates two lightning bolts: one between
null1 and null3, another between null2
and null4

null1 null3 null4 Creates two lightning bolts that share a

null1_dup common source point.

Special Case for Shader Developers Using IRIX 6.x

The mental ray 1.9 renderer and the shaderball previewer both allow
dynamic compilation of shaders, which is useful for testing purposes
when developing a shader. In this case, the shaderball previewer or
mental ray compiles the shader code “on the fly,” and then links in
the resulting object file dynamically.
Note that the shaderball peview is not available with mental ray 2.1.
In this process, the shader code passes through the C compiler and
the default compilation flags used are determined by the operating
system. In IRIX 5.3, the default compilation flag is -32, which
produces object files only usable on mips1 and mips2 executables. In
IRIX 6.x, the default compilation flags used are -n32 (for mips3
executables) and -64 (for mips4 executables).
You cannot mix MIPS versions of executables; for example, if you
load a mips2 shader into a mips3 or 4 shaderball previewer, this leads
to unpredictable results in the shader.
To avoid problems, do the following
• If you are on IRIX 6.x and you are using a MIPS1 or MIPS2 version
of mental ray or the shaderball previewer and you want to use
dynamic compilation, you must set the SGI_ABI environment
variable as follows:
setenv SGI_ABI -32

This instructs the C compiler to compile MIPS1 or MIPS2 object files

for your shader instead of the default MIPS3 or MIPS4 object files.
• If you are on IRIX 6.x and using a MIPS 3 or 4 version of mental ray or
the shaderball previewer and you want to use dynamic compilation,
make sure the above variable is not set. If it is set, unset it as follows:

82 Rendering SOFTIMAGE|3D
 Using Shaders

unsetenv SGI_ABI

SOFTIMAGE|3D Rendering 83
Rendering with mental ray Software

Raytracing Using mental ray

Raytracing calculates the light rays that are reflected, refracted, and
obstructed by surfaces. When rendering with raytracing, you will
have more realistic and precise results. However, it takes much longer
to render with a higher raytracing value.
If you have a scene with reflection, refraction, transparency, or
shadows in it, it will take a fairly long time to render. The beauty of
raytracing is that you can render what you want: if you want to see
only the effects of the animation, or just the effects of reflection, you
can deselect all of the other features and show only these options.
This will take less time to render, yet you benefit from the realistic
results that rendering with raytracing offers.
If there is no reflection or refraction in your scene, then rendering
occurs at about the same speed as hardware rendering.
Each refraction or reflection of a ray creates a new branch of that ray
when it bounces off a solid object and is immediately cast in another
direction. Each new branch can be thought of as a layer: if you add
together the total number of a ray’s layers, it represents the depth of
that ray.
To use raytracing with mental ray, follow these steps:
1. Select an object. If it has some reflection and transparency, the
effects of raytracing are more obvious.
2. Choose the Render menu command. The Render Setup dialog
box is displayed.
3. Select mental ray as the Rendering type.
4. Click the Options button. In the Options dialog box, specify the
Ray Depth parameters.
The Ray Depth option allows you to define the depth of a ray, which
means you are actually defining the maximum number of times a ray
can be reflected or refracted in the scene. The various branches of
rays in a scene constitute a ray tree.
• Reflected ray depth allows you to specify the maximum number of
a ray’s reflective branches in a scene. For example, in a totally
reflective scene, the ray continuously bounces around in the scene
creating an infinite number of branches. This option lets you set an
upper limit on these calculations.

84 Rendering SOFTIMAGE|3D
 Raytracing Using mental ray

• Refracted ray depth allows you to set the maximum number of

times a ray can be refracted in a scene.
• Sum adds the total number of a ray’s reflections and refractions in
the scene. If the total exceeds the number you have specified, the ray
is not cast.
If the Sum exceeds the value specified, the ray is not cast. If the sum
of the two numbers is an odd number, there will be more reflection
rays because reflection always gets calculated first.
5. Click Accept when you have set the values and then render the scene.

Acceleration Method
When choosing an acceleration method for using raytracing with
mental ray, it is wise to return to the findings that you made when you
initially evaluated your scene. The three choices are BSP tree, Ray
classification (mental ray 1.9 only), and Grid (mental ray 2.1 only).
A general rule for choosing between these methods is the overall
complexity of the scene. If the scene information shows that there are
less than 150,000 triangles, then the acceleration method should be
BSP tree (Binary Space Partitioning).

BSP Tree
With BSP tree, the scene is divided into cubes to reduce the number
of computations. Click the Set button beside this option to open the
BSP tree setup dialog box in which you can set the maximum depth
and size.

A Maximum depth can be set to accelerate the renderer by limiting

the level of space subdivision that occurs. The default for maximum
depth is 24.

SOFTIMAGE|3D Rendering 85
Rendering with mental ray Software

In conjunction with the maximum size of the BSP tree, the rendering
process can be adjusted more finely to limit the number of triangles
calculated in each cube by setting the Maximum size. The default for
the size is 4 (triangles).

Ray Classification
Note that Ray classification is available with the mental ray 1.9
rendering software only.
The other method of acceleration, Ray classification, deals with larger
scenes more efficiently than BSP tree. It provides better management
of memory consumption and can help avoid massive memory
swapping. Ray classification exploits the coherence of the rays.
Ray classification checks for intersections between every solid object
to define refraction or reflection.
Click the Setup button beside this option to open the Ray
classification setup dialog box. In it, there are three variables of
optimization to increase the speed of the ray classification.

The Visible option specifies the number of divisions a ray will make
in order to accurately describe the object’s surface.
The Shadow option allows you to specify the shadow ray space division.
The Memory option is a safety device that allows you to specify the
maximum amount of memory to be used for the data structures. The
default value is 10 megabytes.

Grid Acceleration
mental ray 2.1 includes an acceleration method called Grid. The Grid
acceleration method, like other mental ray rendering acceleration
methods, places the scene within a bounding box. The bounding box
size simply represents the volume that objects in the scene occupy.

86 Rendering SOFTIMAGE|3D
 Raytracing Using mental ray

The Grid acceleration method partitions the bounding box into

equal-sized boxes (voxels). This is different from BSP acceleration,
for example, which will create voxels of varying sizes to ensure that
there are an equal number of triangles within each of them. Both the
Grid and BSP methods discard voxels that do not contain triangles.
Only voxels that contain geometry are then evaluated for where the
ray tracer intersects with geometry.
The Grid method works well with scenes that have an even
distribution of triangles in them, since the voxels are not pre-
evaluated for triangles. This also means that the Grid acceleration
method should place fewer demands on memory. Scenes that use
caustics and global illumination are also good choices because they
are typically used in a closed environment. Scenes that are unsuitable
for Grid acceleration are: scenes that do not have an even
distribution of objects, scenes that have an uneven complexity, or
scenes in which caustics or global illumination are not active.
You can optionally change the Grid size by clicking the Settings
button. The Grid size is measured in Softimage units. The size of the
scene’s bounding box (measured by the placement of objects on the
xyz plane) determines the size of the grid you should use.

For example, if you have a few small objects within a large bounding
area (in other words, the objects are far away from each other), you
should specify a smaller grid size. mental ray will have to evaluate
more voxels, but the ray tracer will eventually evaluate fewer voxels.
On the other hand, if you have a large object or a scene full of objects,
you should increase the grid size, as there will likely be geometry in
many of the voxels.

SOFTIMAGE|3D Rendering 87
Rendering with mental ray Software

Antialiasing Using mental ray

Aliasing usually occurs when there is limited pixel resolution.
Antialiasing is a method of smoothing out and sharpening rough or
jagged edges of images to produce a more polished look. It uses a
mathematical process that subsamples the pixel area.
The number of pixels in a scene depends on the screen resolution.
The greater the resolution, the greater the number of pixels. The
smaller the resolution, the smaller the number of pixels.
The more pixels there are, less aliasing occurs and the edge is
smoother. When there are not enough pixels, you need to use
antialiasing to make the lines look smoother.
To use antialiasing with mental ray, follow these steps:
1. Choose the Render menu command. The Render Setup dialog
box is displayed.
2. Select mental ray as the Rendering type.
3. Select the Antialiasing option to open the mental ray Antialiasing
dialog box. When the Active option is selected, the parameters
affect the rendering of the scene.

4. Set the Min Samples and Max Samples values. This describes the
number of samples taken to compare surrounding colour values
which are averaged to define the colour of a pixel. The default
settings for Min Samples and Max Samples are -1 and 1,
respectively.
If the minimum value is zero, the pixel is sampled at least once. The
default minimum of -1 means that the picture will be subsampled
every four pixels (a 2 pixel-by-2 pixel square).
If the default sampling is used, what happens? mental ray takes two
samples, one on pixel 1 and one on pixel 3, compares them and if the
difference is greater than the contrast specified, it tries advancing the
level of samples. Now the new sampling examines pixel 1, pixel 2,
and pixel 3, looking again at the contrast level.

88 Rendering SOFTIMAGE|3D
 Antialiasing Using mental ray

If the contrast is still too high between pixels, mental ray does a third
pass dividing each pixel into four subpixels within itself.
If this still does not meet the supersampling threshold, aliasing
appears because the limit of three levels of sampling has been
executed as the maximum. If this is the case and more antialiasing is
required, you must change the minimum or maximum values to
modify the limit.
5. Set the Threshold Adaptive Sampling sliders. These describe the
contrast level variable for deciding if another level of sampling
will occur.

The smaller the value for each colour (R, G, B), the greater amount
of sampling that must be done to close the contrast gap between the
rendered image and the threshold settings, and the smoother the
antialiasing.
6. Select the Filter Type. The sampling procedure can be
complemented in post-processing to ensure even more
antialiasing. Each of the three filter types (Gaussian, Box, and
Triangle) process subsamples, surrounding and including the
pixel which is being rendered, by using the height and width of
the filter. Based on the value of the pixel, mental ray takes the
average of every pixel and its surrounding pixels and removes
aliasing artifacts.

SOFTIMAGE|3D Rendering 89
Rendering with mental ray Software

The Width and Height options for each of these types define the size
of the filter to be applied.
When you do select a filter, it is applied as an algorithm defining a
curve, peaking at the centre of the pixel sampled.
• The Gaussian filter uses a sloped curve weighting the sampling gently
at the top of the peak and towards the edge of the sampled area.
• The Box filter sums up all the samples in the filter area with an
equal weight.
• The Triangle filter uses a linear curve which affects the pixels by the
least filtering happening at the edges of the sampled area.
filter filter filter
weight weight weight
1 1 1

sample sample sample

sample
distance distance distance
distanc
-size/2 0 size/2 -size/2 0 size/2 -size/2 0 size/2

Gaussian Box Triangle

90 Rendering SOFTIMAGE|3D
 Blurring a Moving Object Using mental ray

Blurring a Moving Object Using mental ray

Motion blur can be used for special effects and on fast moving
objects with lateral motion moving away from to toward the camera.
Motion blur with mental ray uses raytracing to correctly blur
attributes applied to moving objects such as highlights, shadows, and
reflections as well as refractions and intersecting objects.
Because you can apply a motion blur to a selected object in your
scene, it can make rendering go much quicker than when using the
SOFTIMAGE renderer.
Motion blur must be set up for the selected object in both the mental
ray Render Setup dialog box (accessed from the Info > Selection
command) and the Render Setup dialog box (accessed from the
Render menu command) for the effect to work.

3D Motion Blur
To use mental ray on specific objects in a scene, follow these steps:
1. Select an object or objects in a scene.
2. Choose Info > Selection for a single object or Info > mental ray
for multiple objects (click the Render Setup button in the Info
dialog box).
3. Select the Motion Blur option in the appropriate dialog box:
Linear or Exact.

If you want the selected object to have a motion blur based only on a
translation keyframe, select the Linear option. If the object is
animated only in translation, using Linear can actually be an
optimization since only one motion vertex is used for the whole
object. Linear is mandatory for objects whose topology is not
consistent throughout the animation. This affects objects such as
meta-clay and animated Booleans.
If the desired effect requires motion blur by animation of rotation,
scale, cluster, shape, lattice, or materials, select the Exact option.

SOFTIMAGE|3D Rendering 91
Rendering with mental ray Software

4. Click Ok to exit the dialog boxes.

5. Choose the Render menu command. The Render Setup dialog
box is displayed.
6. Select mental ray as the Rendering type.
7. Select the Motion blur option. The mental ray Motion Blur
dialog box is displayed.
8. Select the Active option to activate motion blur, then set the
shutter speed value. Shutter Speed allows you to set the length of
time the shutter is open. A value of zero automatically turns all
motion blur calculations off. The default value is 0.7, with the
range of acceptable values between 0 and 1. The value is in time,
emulating the shutter on a camera being open for a percentage of
time between frame x and x + 1. A large number means a slower
shutter speed.

Controlling the Quality of mental ray Motion Blur

With mental ray 2.1 you can control the quality of the motion blur,
which means that you can decrease the time it takes to render motion
blur scenes.

92 Rendering SOFTIMAGE|3D
 Blurring a Moving Object Using mental ray

mental ray motion blur at lowest mental ray motion blur at medium
quality quality

mental ray motion blur at highest quality

Note: The Motion Blur Quality slider corresponds to the Time

Note
Contrast parameter in the .mi2 file.

1. Choose Render from the Matter module.

2. Select mental ray 2.x as the Rendering Type.
3. Click the Motion Blur button.

SOFTIMAGE|3D Rendering 93
Rendering with mental ray Software

4. Set the Motion Blur Quality sliders. Valid ranges for the quality
sliders are 0 (full quality) and 1 (worst quality). The default
setting is 0.2. Typically, all sliders will be set to the same value.

Motion Blur, Antialiasing, and Rendering Time

Because you can control motion blur quality if you use mental ray
2.1, you may find that you don’t have to increase antialiasing to get
the look you want. This may well save you rendering time, because
motion blur is a purely local effect, whereas antialiasing affects the
quality of the entire rendered image.

Example
A few modifications have to be make to any object that you want to have
motion blur. This is an example of blurring two primitive objects:
1. Get a default Dodecahedron and an Icosahedron.
2. Select the Dodecahedron and choose Boolean > Static. Click Ok
in the dialog box to confirm the other parameters. Select the
Icosahedron to create a union between these two objects.
3. The Boolean > Static command creates a completely new object,
so, you must hide the original objects by choosing Display >
Hide > Unselected.
4. Choose Polygon > Automatic Colourize in the Matter module.
5. Choose the Material menu command and change the shading
model to Lambert.
6. Select the new “jewel” object and choose Info > Selection to open
the Polygon Info dialog box.
7. Change the Automatic Discontinuity value to 44, then click the
Render Setup button and change the Motion Blur option to
Exact. Click Ok to accept the set parameters in the mental ray
Render Setup dialog box and click Ok again to exit the Polygon
Info dialog box.
8. Choose the Render menu command. In the Render Setup dialog
box, select mental ray as the Rendering type, and then select
Motion Blur. In the mental ray Motion Blur dialog box, select
Active and set the Shutter Speed value to 1.
9. Click Accept to accept these changes and exit the Render Setup
dialog box.

94 Rendering SOFTIMAGE|3D
 Blurring a Moving Object Using mental ray

10. In the playback box, change the Start frame to 0 and the End
frame to 30.
11. Spin the object: save a rotation keyframe at frame 0 using
SaveKey > Object > Rotation > All. Then go to frame 30, rotate
the object 180 degrees and save another keyframe.
12. Choose the Render menu command again. Set a resolution of at
least 200 pixels.
13. Set the mental ray Antialiasing Max Samples to at least 3. Lesser
values will have a coarse result and probably won’t look very realistic.
14. Render the sequence using mental ray.
Tip:
Tip To get rid of the noisy look, increase the antialiasing level
and/or add a Blur function from the Antialiasing menu in
the Render Setup dialog box.

2D Motion Blur
In addition to the raytraced (3D) motion blur available for the
mental ray renderer, you can activate post-processing (2D) mental
ray motion blur from within SOFTIMAGE|3D. This feature is like
the mb standalone, which is also a post-processing effect (although
you can still use it with this version).
The 2D mental ray motion blur is much faster than the mb
standalone, although they produce similar results. Depending on the
effect and quality you want to put into your scene, 2D motion blur
may be a solution. At the very least, you can use 2D motion blur to
quickly render your scene, and then use 3D (raytraced) motion blur
for your final render.

How Does 2D Motion Blur Work?

2D motion blur blurs any object that is animated in your scene. This
includes a moving camera, which blurs the entire scene. The blur is
applied based on the shutter speed you specify. This “paints” the blur
onto the rendered scene based on the object’s direction and the
camera’s shutter speed. The mental ray renderer creates a motion
buffer in which a motion vector is stored for each pixel. A motion
vector is the motion (relative to the camera) of the geometry seen
within a particular pixel. Each pixel in the frame is blurred by 2D
motion blur according to the length and direction of the
corresponding motion vector.

SOFTIMAGE|3D Rendering 95
Rendering with mental ray Software

2D motion blur is not of the same quality as true raytraced (3D)

motion blur, but renders much more quickly. If you’re blurring the
entire scene, you’ll find 2D motion blur much faster than raytraced
mental ray motion blur.
This command uses a lens shader and an output shader to produce
the blur effect, which you see when you render the scene. You cannot,
however, edit these shaders.

Limitations of 2D Motion Blur

Because 2D motion blur is not raytraced, certain situations may
cause artifacts in the effect:
• Scenes with reflections, shadows, highlights, or surfaces seen
through transparent objects: 2D motion blur may not create
accurate blurs under these conditions.
• Rotations with change of scale: Artifacts may appear because the
interpolation between motion vectors may cause the background to
show through.
• Obscured backgrounds: When the object passes in front of the
background, the blur only contains colour from the object itself.
• Antialiased or transparent objects: Artifacts may appear if the
object you are blurring is transparent or antialiased. This occurs
because a pixel may be composed of more than one colour, which
may confuse the motion vector painting process.
• Objects whose animation enters or leaves a scene while in motion:
Motion blur may not appear at the beginning or at the end of a
scene because the renderer doesn’t know that the object is in
motion until it enters the scene.

Using 2D Motion Blur

1. Select the object you want to blur and choose Info > Selection.
2. Click Render Setup and select the type of motion blur you want
on the object (Linear or Exact).
3. Choose the Render menu command in the Matter module.
4. In the Render Setup dialog box, select mental ray as the
Rendering Type.
5. Select Motion blur.

96 Rendering SOFTIMAGE|3D
 Blurring a Moving Object Using mental ray

6. In the mental ray Motion Blur dialog box, select the Active option.

7. Select the 2D option. When you activate 2D motion blur, you can
control the way the blur will look by selecting one of the
following blur types:
- Uniform: There is no decay in the blur. The blur’s “trail” is constant.
- Linear Decay: There is decay in the blur, and its gradience is linear.
- Gaussian Decay: There is decay in the blur, and its gradience is
Gaussian, which means that the blur has a contour that peaks at
the centre and ramps off smoothly on both sides of the blur. This
option produces the most visually accurate blur, although it is
more subtle than Uniform or Linear.
8. You can increase the quality of the motion blur by increasing the
number of Samples. Use the default setting of 1 at first to see the
effect, and increase it moderately if necessary. Do not use a
setting of less than 0 because 2D motion blur requires a setting of
a least 1 to produce smooth blur effects.
Tip:
Tip If your object is moving slowly toward the camera, we
recommend a Sample setting of 1, which will give you a
smooth blur.

SOFTIMAGE|3D Rendering 97
Rendering with mental ray Software

Exporting Motion Vectors

You can output the motion vectors that are used to calculate 2D
motion blur to an .mi file without rendering the motion blur itself.
This might be useful if, for example, a shader you have written
requires motion vector data.
To do so, choose the Render command to open the Render Setup
dialog box. Select mental ray as the Rendering Type and then select
the Motion Vectors Only option in the Settings group box.

98 Rendering SOFTIMAGE|3D
 Creating mental ray Shadow Maps

Creating mental ray Shadow Maps

mental ray shadow map at map mental ray shadow map at default
resolution of 125 map resolution of 500

mental ray shadow map at default map resolution

of 500 with antialiasing (0,2)

Note that mental ray shadow maps are only available with the mental
ray 2.1 rendering software.
From Programming mental ray:
Shadow mapping is a technique that generates fast approximate
shadows. It can be used for fast previewing of models or as an
alternative to the more accurate (but also more costly) ray-tracing–
based approach in scenes where accurate shadows are not required.
Shadow maps are particularly efficient when a scene is rendered several
times without changes in the shadows (for example, an animation
where only the camera is moving).
A shadow map is a fast depth buffer rendering of the model as seen from
a light source. This means that each pixel in a shadow map contains
information about the distance to the nearest object in the model in a
particular direction from the light source. This information can be used
to compute shadows without using shadow rays to test for occluding
objects. The shadow computation is based only on the depth
information available in the shadow maps. For fast previewing of
scenes, shadow maps can be used in combination with scanline

SOFTIMAGE|3D Rendering 99
Rendering with mental ray Software

rendering to produce fast approximate renderings with shadows —

without using any ray tracing.
Two different kind of shadows can be produced with shadow maps:
sharp and soft (blurry) shadows. Sharp shadows are very fast, and,
depending on the resolution of the shadow map, they will approximate
the result produced with simple ray tracing. Soft shadows are produced
by distributing one or more samples in a region of the shadow map. This
technique produces soft shadows everywhere and is not as accurate as
the ray-tracing–based approach for computing soft shadows, but it is
much faster.

Creating mental ray Shadow Maps

You create mental ray shadow maps the same way you create
Softimage depth maps. Make sure you’ve selected mental ray 2.1 as
the default previewer/renderer.
Note: If the object casting the shadow is transparent, the object’s
Note
transparency will not be considered when the shadow is
calculated. The shadow will appear as though the casting
object is completely opaque.
1. Create a spot light.
2. Select Depth Map in the Define Light dialog box.
3. You can control the shadow’s resolution using the Map
Resolution parameter, which determines the number of pixels
used in the shadow map. The greater the number of pixels, the
better the result will be, but the longer the rendering time will be!
A good rule of thumb to begin with is to set the depth map
resolution to half of the resolution of the rendered image.
Alternatively, mental images recommends starting with a
resolution of 500 pixels.

100 Rendering SOFTIMAGE|3D

 Creating mental ray Shadow Maps

Creating Soft Shadow Maps

mental ray soft shadow map at mental ray soft shadow map at default
default map resolution of 500 map resolution of 500 with antialiasing
(0,2) and adjusted step filter and size
values of 50 and 15

You can create softer mental ray shadow maps the same way you
create Softimage soft shadows, except that the Penumbra parameter
is not used by mental ray. Instead, the Filter Size parameter
determines the softness of the shadow’s edges.
Note: You shouldn’t mix area lights and mental ray mental ray
Note
shadow maps, since this combination will likely produce
artifacts. The mental ray shadow map will be jittered by the
area light, which will produce incorrect depth information
and put shadows on objects that in fact cast shadows.
1. Create a spot light.
2. Select Soft in the Define Light dialog box.
3. You can use the Umbra Intensity parameter to define the
intensity of the shadow (0 = black shadow, 1 = no shadow). Try
starting with an intensity of 0.2.
4. You can control the shadow’s resolution using the Map
Resolution parameter in conjunction with the Filter Size and
Filter Step parameters. For a definition of the Map Resolution
parameter, see step 3 in the previous section.
- Filter Size determines the size of the box filter, which is used to
soften the shadow’s edges. A good rule of thumb is to set the
number to about 10% of the Map Resolution. This parameter
identifies the amount of softness in the shadow.
- Filter Step determines the pixel offset, which is used to apply the
filter at a specific number of pixels from the previous application.
This parameter identifies the quality of the softness. The greater
the step, the better the result will be, and (of course) the longer

SOFTIMAGE|3D Rendering 101

Rendering with mental ray Software

the render will be. A good rule of thumb here is to add 3 steps for
every 10% of the Filter size.

Shadows Maps and Antialiasing

You can use antialiasing to improve the look of the shadow map.

Troubleshooting
Problems can occur with shadow maps when the object casting the
shadow touches the shadow; the rendered image may incorrectly show
the shadow starting on the object or starting a pixel or two away from
the object. To fix this problem, increase the Map Resolution.

102 Rendering SOFTIMAGE|3D

 Creating mental ray Displacement Maps

Creating mental ray Displacement Maps

Note that mental ray displacement maps for polygonal objects are
only available with the mental ray 2.1 rendering software.
Unlike bump mapping, displacement mapping actually introduces
new vertices and triangles so that during rendering the object’s
geometry is altered to create a bumpy surface. The displaced surface
is visible in shadows.
Note, however, that displacement mapping requires more memory than
bump mapping, so rendering time will be increased if you use them.
You can apply displacement maps to the entire object or to certain
polygons (if it’s a polygonal objects) using local textures.

Applying Displacement Maps to Polygonal Objects

Avoid using textures with fine details—rendering time may suffer as a
result of trying to replicate the details of the map in the mesh’s geometry.
You can control the amount of new geometry mental ray creates
using the Surface Approximation settings (see page 104). If you don’t
enable Surface Approximation, mental ray will only displace existing
vertices, which will likely be insufficient to create the look you want.
1. Select the object to which you want to apply a displacement map.
2. Select Texture > 2DLocal or 2DGlobal.
3. Enter the name of the texture you’re using to create the
displacement map in the Picture Filename text box.
Note: If you use an image that contains a lot of brightness (with
Note
little or no black or dark areas), the polygons will be pushed
away from their original position.
4. Choose a mapping method that conforms well to the shape of the
polygon mesh you want to displace.
5. Select the Displacement (mr) option and set the Roughness
slider to a fraction of 1 (such as 0.1 or 0.01) or greater. A negative
value displaces the surface inward.

SOFTIMAGE|3D Rendering 103

Rendering with mental ray Software

Controlling Displacement Map Quality with Surface

Approximation
Surface approximation defines how an object will be tessellated when
it’s rendered. Surface approximation is now available for polygons so
that you can control the quality of the displacement that occurs when
the mental ray 2.1 renderer creates the displacement.
1. Select the object to which you want to apply surface
approximation.
2. Choose Info > Selection.
3. Click Render Setup.

4. Select Surface Approximation.

5. You can choose between two tessellation methods (Parametric is
not available for polygonal objects):
- Spatial determines the maximum triangle size. For example, if
you keep the default of five Softimage units, only triangles that
exceed this size will be subdivided. For the mesh surface to
conform to the information in the texture, it’s likely that you’ll
want to decrease this value. You can set Subdivision limits to
control the number of times the surface is subdivided.

104 Rendering SOFTIMAGE|3D

 Creating mental ray Displacement Maps

- Curvature tests the curvature (angle) between two triangles. If

the angle is greater than the one provided, the triangle is
subdivided again. It calculates enough subdivisions to create
refined curves, but only creates triangles where the curvature
demands it. You can set Subdivision limits to control the number
of times the surface is subdivided.
- Units definition determines whether the triangles are measured
in pixels or Softimage units. Pixel length sets the units to
represent diagonal pixel length. This means that no surface
triangle will be larger than the pixel length you specify. This is
used to guarantee the accuracy of the triangles representing the
surface. System sets the units to Softimage units.
6. Click Ok.

Troubleshooting: A Word about Normals

The mesh’s normals are used to calculate the displacement. If you’re
getting unusual results, display the object’s normals (Show >
Normals). Normals for most objects created in SOFTIMAGE|3D
should be one Softimage unit in length. If you see normals longer
than this, scale the object up in size, freeze its scaling (Effect > Freeze
> Scaling from the Model module), and then choose Effect >
Recompute Normal from the Model module.

SOFTIMAGE|3D Rendering 105

Rendering with mental ray Software

Rendering Faces of an Object

By default, only the front faces of an object are rendered; that is, the
ones whose normals are facing the camera. The back faces are
“culled” (ignored). If you like, you can render all faces of an object.
This could be useful if you are rendering something like a ribbon or
pages in a book, in which case you would want both sides of the
object to be rendered.
In mental ray, you have the option of rendering the Front, Back, or
Both faces. The default is Both.
1. Choose the Render menu command.The Render Setup dialog
box is displayed.
2. Select mental ray as the Rendering Type.
3. Click the Options button. The mental ray Options dialog box appears.
4. Selecting a Face allows you to specify which polygons are to be
rendered. For example, if you select Front only the polygons
facing the camera are rendered.

If the back faces of the object are not influential in the scene, then
you could select only the Front face option. However, to ensure
the highest quality image, perhaps for the final render, you
should select the Both option. You would also select Both is you
are rendering a grid (such as a flag) where you need to see both
sides of the object.

106 Rendering SOFTIMAGE|3D

 Rendering Shadows

Rendering Shadows
For render tests, you can turn off all reflectivity, refraction, or
textures. You can make sure that no shadows are cast for objects. It is
advantageous to be able to turn off shadow components because they
are time-consuming to render. When you’re ready to see the
reflectivity, refraction, textures, or shadows again, you can select
them to make your scene more realistic looking.
1. Choose the Render menu command. The Render Setup dialog
box is displayed.
2. Select mental ray as the Rendering type.
3. Click the Options button. The mental ray Options dialog box appears.
4. Select Trace to activate the raytracer and/or Shadow to create shadows.
The raytracer is automatically selected if a lens shader is active.
A set of global switches for raytracing parameters is also located here.

Note: For information about creating shadow objects with mental

Note
ray, see Creating Shadow Objects on page 40 in the Defining
Materials and Textures User’s Guide.

SOFTIMAGE|3D Rendering 107

Rendering with mental ray Software

Surface Approximation
It is possible to reduce the number of triangles in the geometry of an
object and still render a very smooth surface. Dealing with large
models while using parallel processing can slow down the time to
distribute the information as well as occupy unnecessary swap space.
For information on using Surface Approximation in conjunction
with displacement maps, see Controlling Displacement Map Quality
with Surface Approximation on page 104.
To set the surface definition for an object:
1. Select the object.
2. Choose the Info > Selection command and click the Render
Setup button.
3. In the mental ray Render Setup dialog box that appears, you can
decrease the step size, and the number of triangles decreases
drastically.
4. Select a type of surface approximation: Static or Adaptive (see
the next section for more information).
Depending on the surface approximation you choose, mental ray
tries to create the perfect curve between two points.
5. Set the Subdivision limits.
To prevent massive amounts of subdividing, a limit can be assigned
to each selected object which helps to optimize the quality of surfaces
with areas of different curvatures.
There can be a minimum (Min) amount required guaranteeing a
smooth curve, and a maximum (Max) amount guaranteeing the
utmost limit of computation.
6. Click Ok in this dialog box and the next to save your changes.

108 Rendering SOFTIMAGE|3D

 Surface Approximation

Choosing an Approximation Method

Static
The Static Parametric method of surface approximation defines the
U and V steps of the selected object as seen by the geometry.

Changing these values also change the number of steps in the object’s
Info selection dialog box. These parameters remain the same
throughout the animation and if the camera gets very close to the
object, the steps might be seen.
This is the same as the method of surface approximation used by the
SOFTIMAGE renderer.

Adaptive
The Adaptive method of surface approximation interprets the size of
the object in proximity to the camera and adapts the number of steps
to account for the size in the render.

Angle Patch edge

Division

Distance specified by Adaptive

options (Chord length)

This is the best method for surface approximation for both NURBS
and patch surfaces.
The size of the step can be measured in either the number of pixels in
the length of the step or the SOFTIMAGE|3D units, defined by the
grid setup.
Adaptive surface approximation can be accomplished by one of two
methods: Spatial or Curvature.

SOFTIMAGE|3D Rendering 109

Rendering with mental ray Software

Spatial
The Spatial surface method using the pixel length is more intuitive
than using the system units for definition of the step’s length. Within
the object curve, the steps are as long as the number of pixels
specified in the text box.

Pixel Length
of Step Patch edge

Division

Distance specified
by Adaptive options
(Chord length)

If the step is longer than the specified number of units, the steps are
subdivided until they are equal to that number.

Curvature
The Curvature method chord length uses the distance between the
most distant arc of a curve and the step as a parameter by which the
steps will be subdivided. This method is more effective when dealing
with larger flat areas which converge into curved surfaces.
It approximates the flat area with less tessellation than Spatial and
approximates the curved area with more subdivisions of steps where
necessary.

110 Rendering SOFTIMAGE|3D

 Surface Approximation

If the combination of the Chord Length and the Angle of the steps do
not meet the parameters in the text boxes, the steps are subdivided.

SOFTIMAGE|3D Rendering 111

Rendering with mental ray Software

Using Output Shaders

Output shaders operate on rendered images before they are written
to a file, but after the normal rendering process is complete. They can
perform operations such as filtering, compositing with other files,
and writing to different file formats. The benefits of using output
shaders is that the rendering time is usually very quick and can
produce excellent results when used correctly. However, they have
some limitations because they are simply added to the existing
rendered image.
Before you choose an output shader, you can set up these options:
Contour Rendering, the Surface Normals, and Z pic information.
These options are available when you:
1. Choose the Render menu command.
2. Select mental ray as the Rendering type.
3. Click the Options button.
The parameters are in the Output Shaders area of the mental ray
Options dialog box.

Setting Up to Use the Output Shaders

Using Contour Rendering
Note that contour rendering is not available with mental ray 2.1.
This method of rendering defines surfaces by contrast rather than by
geometry and produces antialiased contour lines between surfaces as
well as in areas where surface normals show discontinuity. Contour
line rendering is particularly useful for cartoon animation because
the contour lines that are produced look similar to line drawings of
traditional animation, and the output can be painted cel by cel.
When you click the OFF/ON button beside the Contour rendering
option, the Contour Rendering Parameters dialog box appears:
• Selecting Activate makes the Contour Rendering option active.
• Line Width defines the width of the contour lines in pixels. This is
0.5 by default.
• Contour Depth is used to give a tolerance to distinguish between
objects that are almost in the same position in space. This is 0 by
default, which means it will distinguish between all objects.

112 Rendering SOFTIMAGE|3D

 Using Output Shaders

Display is disabled during contour rendering, with the result written

to a .pic file. This is the only output file generated.
Note: You cannot use contour rendering with the BSP tree
Note
acceleration method. If you do so, SOFTIMAGE|3D
switches the Ray Classification method instead to render the
contour images.

Surface Normals
The Surface Normals option activates surface normal output and
creates an output file with the .n extension. It will have the same
resolution as the actual picture, except that instead of storing the
RGB value for each pixel, it stores the surface normal used for
rendering the triangle nearest the camera. Normal encoding is used
mainly for post-processing applications.

Z pic
The Z pic option activates depth picture output and creates an
output file with the extension .Zpic. This method is similar to
Normal encoding, except that it is the distance between the camera
and the nearest triangle that is stored instead of the RGB value.

Selecting an Output Shader

If you need to search for an output shader, click the Select button
beside the Output shader scroll box and choose a shader from the
appropriate chapter or database in the browser that appears.

SOFTIMAGE|3D Rendering 113

Rendering with mental ray Software

• Edit displays the dialog box containing the parameters that define
the shader you have selected. You can then edit the parameters of
that shader. For example, if you are using the same shader two or
more times in the list, you can modify the parameters slightly for
each time it is used.
• Active makes the shader active or inactive, depending on the
current state of the shader. If a shader is active, it is highlighted.
• Move Up lets you rearrange the order of the shaders when you have
more than one active shader in the list. Since you can have more
than one lens shader active at the same time, it makes a difference
how you order them in the list. The shaders are processed starting
from the top of the list.
• Delete deletes the currently selected shader from the shader list.
• Set Name lets you change the name of a output shader. To change
names, select a shader from the list, modify the name in the Name
text box, and click Set Name. This creates a new shader with the
new name, but with the current parameters. You can edit these
parameters in the dialog box that appears when you click Edit (see
the previous description of Edit). The new shader can then be saved
and recalled by name in other scenes.
• The Name text box lets you save a shader by another name (see Set
Name above).

Example
1. Select an object or load a scene.
2. Choose the Render menu command. In the Render Setup dialog
box, select mental ray as the Rendering Type and set the Start and
End frames 1.
3. Click the Options button. In the Output Shaders area of the
mental ray Options dialog box, click Select.
4. In the browser that appears, go to the /Shader_Lib/GLOWS
directory. Select DGlow as the Output shader and click Load.
5. With the DGlow shader selected, click Edit.
6. In the OZ-Diffusive Glow Postfilter dialog box, click the Select
button beside the Object list text box. The Object list allows you
to apply the DGlow shader to any or all objects in your scene.
Select the object you want and click Ok to exit the Object list.

114 Rendering SOFTIMAGE|3D

 Using Output Shaders

7. Click Ok to exit the OZ-Diffusive Glow Postfilter dialog box and

then click Accept in the mental ray Options dialog box.
8. Click Accept to exit the Render Setup dialog box.
9. Choose Preview > All to view the results.
10. If you are satisfied, choose the Render menu command. In the
Render Setup dialog box, name your scene appropriately, and
then click Render Sequence to begin the rendering process.
Note: You will notice that mental ray only applies the output
Note
shader to the specified object after it has completely
rendered the scene.
11. To view the final result, choose the Picture menu command in
the Tools module. Use the browser to go to the
RENDER_PICTURES chapter. Select your rendered scene and
then click Display.

SOFTIMAGE|3D Rendering 115

Rendering with mental ray Software

Saving to a File
You can write the commands received by the raytracer to a file
instead of actually rendering the scene.
1. Choose the Render menu command. The Render Setup dialog
box is displayed.
2. Select mental ray as the Rendering type.
3. Click the Options button. The mental ray Options dialog box appears.
4. Select the Output to File option.
5. Enter the complete path name or a relative path name in the text
box; the file will be written to the current working directory. This
output can then be edited and manually sent to the mental ray
renderer using the irix command.
If the Output to File per frame option is not selected, all frame are
created in the same .mi file. By default, all information for the
sequence of output frames are written into a single file. If the Output
to File per frame option is selected, each frame in the sequence is
written to a separate file.

116 Rendering SOFTIMAGE|3D

 Distributed mental ray Rendering between IRIX and Windows NT

Distributed mental ray Rendering

between IRIX and Windows NT
The mental ray rendering executable is able to distribute rendering
tasks to multiple hosts. The machine which starts the rendering
process (the rendering “client”) reads the scene from a text file which
is either read from the disk or piped directly into mental ray by
SOFTIMAGE|3D during interactive rendering. The client then sends
the scene to the other machines that participate in the rendering (the
rendering “slaves”). This works well when all the machines have the
same architecture and all see the file system the same way (either by
having exact duplicates on each machine or by having an NFS
mounted network). However, this situation breaks down when the
machines have different architecture and both the setup and the file
name conventions are different.

Localization and File Naming Problems

The mental ray renderer must reference the shaders’ dynamic
libraries and support files using an absolute path. This information is
passed on to the rendering slaves by the rendering client. All the
rendering slaves have to load the shaders’ dynamic libraries in order
to render their part of the final image.
The problem is two-fold when a mixed Windows NT and IRIX
network is involved for slave rendering: one problem is how each file
system names its absolute paths and another problem is where
SOFTIMAGE|3D is installed so that mental ray can locate the shaders.

File Naming
By convention, all file references that SOFTIMAGE|3D is generating
in the scene description to mental ray (the MI file) follow the IRIX
format. This ensures that all the rendering slaves interpret the path
names the same way. Another reason is that the MI file is platform
independent so that an MI file created using a Windows NT version
of SOFTIMAGE|3D can be used on the IRIX version. The
SOFTIMAGE|3D to mental ray translator also converts the file
extensions to ensure cross-platform compatibility. The standard is in
IRIX format, so the file extensions related to dynamic libraries and
shader compilation are converted as follows:
.dll ¹ .so
.lib ¹ .a
.obj ¹ .o

SOFTIMAGE|3D Rendering 117

Rendering with mental ray Software

File Localization
The correspondence between the shaders’ libraries on Windows NT
and IRIX is made using the linktab.ini file mechanism since both
versions are installed differently on both platforms. The installation
script installs all the related components of SOFTIMAGE|3D in a
single directory tree. This tree is pointed to by the environment
variable SI_LOCATION. The following is the current directory
structure (in part) after installation on a Windows NT machine:
$SI_LOCATION\3D\
\bin
\rsrc
\flexlm
\setup
\mental_ray
\bin
\inc
\lib
\man
\MR_Shaders
The default installation on Windows NT puts this tree in
c:\SI_LOCATION\3D and in /usr/softimage/3D for the IRIX version.

Setup
All the Windows NT machines, the rendering client, and the
rendering slaves which are used as part of the distributed rendering
must have a linktab.ini file located in the directory pointed to by the
SI_LOCATION environment variable.
The linktab.ini file is an ASCII file that is used to define one-to-one
relations between IRIX and Windows NT paths. Each line of the
linktab.ini file represents such a relationship and contains two
entries. The first entry on the left is the Windows NT path, and the
entry on the right is the IRIX equivalent.
On the Windows NT machine, make sure that the SI_LOCATION
environment variable points to the location where SOFTIMAGE|3D
is installed. This variable should be present in the
%SI_LOCATION%\3D\bin\SETENV.BAT file. To modify or see the
contents of this file, click the User Tools option in the
SOFTIMAGE|3D program group.

118 Rendering SOFTIMAGE|3D

 Distributed mental ray Rendering between IRIX and Windows NT

Create a file in the directory which SI_LOCATION points to and

name it linktab.ini. This file contains path correspondence between
Windows NT and IRIX for the current machines. In this file, type the
following commands (each of these should be on one line):
C:/SI_LOCATION/mental_ray/MR_Shaders/
/usr/softimage/mental_ray/MR_Shaders/
C:/SI_LOCATION/3D/bin/rsrc/usr/softimage/3D/rsrc

It is assumed that the installation of SOFTIMAGE|3D is located in

/usr/softimage.

The first line ensures that the shaders’ dynamic libraries are visible on all
the platforms. The second ensures that the correspondence between the
rsrc directory on IRIX and Windows NT, especially that the softimage.mi
and noIcon.pic files are looked up correctly. The noIcon.pic is only used to
replace pictures that are not found during rendering.
Repeat these steps for every Windows NT machine that you wish to
include in the distributed mental ray rendering.

Windows NT Only Networks

The linktab.ini file is not needed on an homogeneous network of
Windows NT machines for which the installation is identical. This
means that each Windows NT machine has SOFTIMAGE|3D
installed in exactly the same location on the same drive. If the drive
or path name is changed on one machine, then all the machines must
have a linktab.ini file similar to the previous sample to map each
other’s installation correctly and thus be able to perform mental ray
distributed rendering.
While the linktab.ini file is not required for a homogeneous network,
its presence does not make a difference to normal operation.

Debugging
On a Windows NT machine, generate an MI file that contains a
default sphere with a material shader on it and examine its contents.
(any material shader will do). All file names should be in IRIX
format. If there is one path that is not in IRIX format, then there is a
typographical error in the linktab.ini file. To generate an MI file,
choose the Render menu command in the Matter module. In the
Render Setup dialog box, select the mental ray renderer and select
Options. The MI file generation controls are located in the lower
right corner of the dialog box.

SOFTIMAGE|3D Rendering 119

Rendering with mental ray Software

Interactive Selection of Hosts

The Info > MentalHost command in the Matter module lets you
interactively select rendering host machines.
This means that you no longer have to iconize SOFTIMAGE|3D and
edit the .rayhost file in your HOME directory to modify host
information and then return to SOFTIMAGE|3D to start the render.
To make this command available, you must use the new mr_cusgen
standalone to define the servers that the users can select at runtime.

Defining Servers Using the mr_cusgen Standalone

The mr_cusgen standalone allows you to define the host servers
available to users through the MentalHost interactive plug-in.
Typically, this task is performed by the system administrator, who
uses the mr_cusgen standalone each time the network changes (for
example, when hosts are added or removed).
After using mr_cusgen, the results are:
• Users can choose the Info > MentalHost command in the Matter
module. In the related dialog box, they can select the host servers to be
used for rendering from among the currently available host servers.
• Upon exiting the dialog box, the .rayhost file in their HOME
directory is automatically updated with this information.
To set up this command:
1. Create an ASCII file listing the names of servers available for
rendering on the network. List one server per line, for example:
host 1
host 2
host 3
2. Separate hosts into groups by placing a line between them that
begins with the hash (#) character. These host names will appear
in the dialog box displayed when users choose Info >
MentalHost in the Matter module.
3. Save this file under any name, such as <filename>.rayhost in
your home directory ($HOME on IRIX and Windows NT).

120 Rendering SOFTIMAGE|3D

 Interactive Selection of Hosts

4. At the command line, type:

mr_cusgen <file name>

using the file name you created in step 3.

This generates a mentalHost.cus file in the directory pointed to by
the SI_CUSTOM_MATTER environment variable. This file contains
the related information, which is global to the network.

Selecting Hosts for Rendering

Before rendering in SOFTIMAGE|3D, you can:
1. Choose the Info > MentalHost command in the Matter module.
The following dialog box is displayed.

2. Select or deselect the server hosts for the rendering job. Your
local .rayhost file is automatically updated so that the rendering
job will use the hosts you selected.
3. Render using the mental ray renderer.

SOFTIMAGE|3D Rendering 121

Rendering with mental ray Software

Enabling mental ray Messages During

Rendering
When rendering using soft -R on Windows NT, errors and warnings
do not appear in the console. If you enable the MR_VERBOSE_FILE
environment variable, the messages will appear in a text file.
For example,
set MR_VERBOSE_FILE=mrerrorlog.txt

creates the text file in the current directory. Different renders are
separated by dashed lines (-----) in the file.
You can also specify a full path for the file name:
set MR_VERBOSE_FILE=C:\users\maggie\mrerrorlog.txt

More verbosity!
mental ray can report more complete information if you specify the
following:
set SI_MI_TRACER2 c:\Softimage\etc\etc\ray2.exe -verbose 6

"-verbose on" is the equivalent of setting verbosity to 5. The

maximum you can specify is 6. Note that this works on IRIX
machines as well.
Also, for distributed renders, in the .ray2hosts you can tell the slave
to output more complete verbose information.
For example,
Matrix -verbose 6
Starwarz -verbose on
Mummy -verbose off
All the slave’s messages will be redirected to master and will be
logged in the file specified with the MR_VERBOSE_FILE
environment variable (Windows NT only).

122 Rendering SOFTIMAGE|3D

 Rendering a Range of Frames

Rendering a Range of Frames

To render a range of frames contained in a mental ray version 2.1 .mi
file, use this syntax:
ray2 -render start end step

• start determines which frame render should start.

• end determines which frame render should stop.
• step specifies how many frames mental ray should skip before
rendering the next.

Example
If the .mi file contains frames 300 to 400 and you want to render
frames 330 to 360, you would enter the following:
ray2 Frame300-400.mi -render 30 60 1

mental ray will start rendering when it reaches the thirtieth frame
contained in the .mi file and will stop when it reaches the sixtieth
frame contained in the file.

SOFTIMAGE|3D Rendering 123

Rendering with mental ray Software

124 Rendering SOFTIMAGE|3D

 Index

Index motion blur 54

rendering 62
Depth of field 9
A Depth of trees 47
Acceleration method in mental ray 85 Depthcue rendering 26
Adaptive antialiasing 50 Displacement mapping 103
mental ray 89 Dithering 45
Alpha channel and motion blur 54
Angle of camera 9 E
Antialiasing 45, 49 Edges, hiding and showing 39
filters
adaptive 50 F
Bartlett 49 Faces, rendering 64, 106
comparisons 50 Field rendering 45, 57
mental ray 88 scripts 58
motion blur 56 Files
Aspect ratio of camera 10 .lin 20
.mi 116
B rayhosts 67
Back culling 64 saving raytracer commands 116
Background, rendering in 32 scripts 34, 58, 81
Bartlett antialiasing filter 49 .Zpic 113
BSP tree in mental ray 85 Filters, antialiasing
adaptive 50
C Bartlett 49
Camera 8 comparisons 50
angle 9 mental ray 89
aspect ratio 10
depth of field 9 G
lens shaders 10 Ghost rendering 28
movement 12
projection 13 H
setting and resetting 8 Half blur 53
Colours with motion blur 56 Hidden-line rendering 26
Command line hiding and showing edges 39
motion blur 53 previewing 20
rendering 31, 37 Hiding edges for rendering 39
Compositing 23
Contour rendering 112 L
Lens shaders 10
D Lights 7
Depth computation optimizing rendering time 23
mental ray 113 .lin files 20

SOFTIMAGE|3D Rendering 125

Index

M colour 56
depth computation 54
Materials first/last frames 55
optimizing rendering time 22 growth and scale 56
previewing 19 half blur 53
mb standalone 53 high curvature movement 55
Memory inaccessible data 56
guidelines 23 limitations 55
motion blur 54 mb standalone 53
requirements 24 memory 54
mental ray renderer 25, 67, 72 mental ray 91
acceleration method 85 rotation 55
adaptive supersampling 89 transparent objects 56
antialiasing 88
BSP tree 85 N
contour rendering 112 Network rendering 67
depth computation 113 Normals 7
displacement mapping 103 mental ray 113
faces 106
motion blur 91 O
multiple objects 69 Optimizing rendering time 22
networks 67 Output shaders 112, 113
optimizing time 23 setting up 112
output shaders 112, 113 surface normals 113
previewing 74
ray classification 86 P
rayhosts file 67 Pre and post-frame scripts 58
raytracing 84 Previewing 19
saving raytracer commands to file 116 .lin files 20
script files 81 mental ray 74
setting up 70 subregions 19
shaders 68, 79, 81 Projection, camera 13
shadows 107
single objects 69 R
surface approximation 108 Ray classification 86
surface normals 113 rayhosts file 67
Z channels 113 Raytracing 46
.mi files 116 acceleration method 85
Modelling, optimizing rendering time 22 mental ray 84
Motion blur 45, 52, 53 reflection mapping 48
alpha channel 54 saving commands to file 116
antialiasing 56 Reflection mapping 48
calculating 53 optimizing rendering time 22

126 Rendering SOFTIMAGE|3D

 Index

Rendering 29 SGI hardware renderer 27

in background 32 Shaders 68, 79
command line 31, 37 lens 10
contours 112 linking to script files 81
depth information 62 output 112
depthcue 26 Shading 7
faces 64, 106 Shadows, rendering 107
field rendering 45, 57 Showing edges for rendering 40
ghost 28
SOFTIMAGE renderer 25
hidden-line, see Hidden-line rendering
Standalones
mental ray, see mental ray renderer
motion blur 53
motion blur 52, 91
optimizing time 22 renderer 31, 37
output shaders 112 Subregions
rotoscope 28 previewing 19
scenes 29 rendering 36
scripts 34 Surface approximation 108
sequences 25 adaptive 109
resuming 30 curvature 110
SGI hardware 27 spatial 110
shadows 107 static 109
SOFTIMAGE 25 Surface normals 7
standalone 31, 37 mental ray 113
subregions 36
surface normals 113 T
tag channels 61 Tag channels 61
wireframe 26 Tessellation 7
Z channels 62 Textures, previewing 19
Resolution 30
Time, optimizing 22
Rotoscope rendering 28
Time/date stamp 30
S Tree depth 47
Saving raytracer commands to file 116 Triangles 7, 47
Scenes U
previewing 19
rendering 29 User tag 61
Scripts W
field rendering 58
linking to shaders 81 Wireframe rendering 26
pre and post frame 58
rendering 34
Z
Sequences, rendering 25 Z channels 62, 113
resuming 30 .Zpic files 113

SOFTIMAGE|3D Rendering 127

Index

128 Rendering SOFTIMAGE|3D