June 95 - QuickDraw 3D: A New Dimension for Macintosh Graphics
QuickDraw 3D: A New Dimension for Macintosh Graphics
Pablo Fernicola and Nick Thompson
QuickDraw 3D is a new technology that helps developers bring 3D capabilities to
their applications. It runs on all Power Macintosh computers and offers
high-performance 3D rendering and other features that make working with 3D data
easier. This article gives the basics you'll need to use QuickDraw 3D in your
application, whether you're a consummate 3D developer, a classic 2D application
developer, or a game developer.
QuickDraw 3D is the newest enhancement to the Macintosh graphics architecture.
Developers have been requesting a 3D library, supported at the system level,
since the Macintosh was introduced. Although a number of Macintosh developers
have produced some amazing 3D applications, 3D graphics capabilities were
relegated to niche applications due to the lack of support at the core
operating system level. QuickDraw 3D, which is expected to ship in mid-1995,
brings the ability to deal with 3D graphics to all Power Macintosh
applications: not only can traditional 3D applications take advantage of it,
but it provides base functionality for general-purpose applications as well.
QuickDraw 3D is a Code Fragment Manager-based shared library, with a C-based
API. Here we'll cover some concepts you need to know to get basic QuickDraw 3D
support into your application. This issue's CD contains a prerelease version of
the QuickDraw 3D shared library, the 3D Viewer shared library, programming
interfaces, preliminary Inside Macintosh: QuickDraw 3D documentation, sample
code, utility libraries, and other goodies. Two of the sample programs are
discussed in this article.
The API described in the article is based on a beta version of QuickDraw 3D;
although nearly final, the API may change before the final release of the
software.
In addition, we'll talk about reading and writing data in QuickDraw 3D metafile
format, which is a way of representing 3D data in a consistent, transferable
manner. But first we'll set the stage with some background information.
QUICKDRAW 3D -- SO, WHAT'S THE BIG DEAL?
As we'll explain further in this article, QuickDraw 3D provides developers with
a number of benefits:
- a rich set of high-level geometries
- built-in renderers that cover the base functionality needed by
developers
- immediate and retained graphics
- a common 3D file format
- human interface guidelines and widgets
- a 3D pointing-device manager that provides support for input devices
with more than two degrees of freedom
- pointing and picking support that enables user selection of 3D data
- transparent access to graphics accelerators
- an extensible, plug-in shading and rendering architecture
- implementation advantages over other 3D libraries
We've made
dealing with 3D data in applications easier with QuickDraw 3D. By creating a
standard for data interchange, with a well-rounded metafile definition, we're
enabling applications to read and write 3D data in a consistent format. The
metafile specification addresses requests from both end users (who couldn't
exchange data between applications in a common format) and developers (who had
to write special-case code to deal with several different file formats).
QuickDraw 3D comes with a set of human interface guidelines to foster the
adoption of a consistent look and feel between applications (see "The QuickDraw
3D Human Interface"). 3D applications today are geared toward the trained 3D
expert; what you learn in one application is generally not transferable to
another application. By following the QuickDraw 3D human interface guidelines,
however, developers can help make 3D graphics an integral part of the user
experience within their applications.
THE QUICKDRAW 3D HUMAN INTERFACE
BY DAN VENOLIA
QuickDraw 3D provides human interface guidelines (in version 1.0) and a toolkit
for implementing the guidelines (to come in the second major release). A sample
application on this issue's CD illustrates our current ideas for a 3D human
interface. By getting a preview of our plans, you can start taking your
applications along the common path.
Our main goal is to provide integration into the Macintosh experience. We feel
that 3D graphics will be the next popular multimedia data type -- in the way
that 2D graphics, sound, and movies have been in the past -- and users will
want to incorporate 3D data into their documents in the same way that they can
now incorporate other multimedia data types. To do this they'll need an
interaction model built on the 2D principles that they're familiar with.
Our guidelines offer suggestions and examples of how things can be done. If
your applications are targeted for a very specific audience, and you know that
audience well, you may decide to communicate with them in a different way, and
that's perfectly OK.
One of our guidelines, about direct manipulation through the use of a widget,
is illustrated in Figure 1. Here we've appropriated the 2D grab handles that
are popular in many "draw" programs and extended them to 3D. A widget is a set
of handles for control of spatial parameters. Some widgets, such as the scale
tool shown in Figure 1, indicate selection of a shape, while others make an
invisible object, such as a light or a camera, visible.
Figure 1. A scaling widget
Figure 2 shows what a full-featured 3D application might look like. The
emphasis here is on what's the same as in 2D applications rather than on what's
unique. The illustration shows a shape selected with a rotation widget, a
material selection palette, a room metaphor, and a document containing multiple
views of a scene.
Figure 2. Conceptual sketch of a 3D application>
QuickDraw 3D technology has been made possible in part by the dramatic
performance improvements in the Power Macintosh line of computers. The
performance of QuickDraw 3D is scalable across the Power Macintosh line; we've
put in a lot of effort to ensure that the performance on even entry-level
computers is excellent. With hardware acceleration, these computers can easily
compete (and win) against mid-range workstations costing a lot more money.
HOW QUICKDRAW 3D COMPARES WITH OTHER LIBRARIES
QuickDraw 3D offers many advantages over other 3D libraries. When using other
graphics libraries, you're on your own if, for instance, you want to change the
way a scene is rendered (say, by doing ray tracing or applying procedural
shading): you have to reimplement all of the 3D architecture. With QuickDraw
3D, you only have to write code to deal with the specific area that you want to
change. And, even better, the code you write can be used as a plug-in by other
applications.
Unlike some libraries, QuickDraw 3D will be able to take advantage of a number
of 3D hardware acceleration solutions, since acceleration was one of its design
criteria. Another important criterion was cross-platform support. For example,
a renderer could be written to take advantage of low-level 3D libraries, such
as the Silicon Graphics OpenGL graphics library.
WHAT YOU CAN DO WITH QUICKDRAW 3D
The 3D application development process can be broken down into four areas:
creation of 3D data into a set of data structures, manipulation of that data in
the human interface of the application, presentation of the data by displaying
it, and transportation of the data (saving to and reading from files).
QuickDraw 3D provides support in each of these areas. You can implement one or
more of them in your application:
- QuickDraw 3D geometries -- If you're planning to write an application to
deal with the creation of models, QuickDraw 3D lets you define the
representation of the objects to be modeled in 3D form.
- QuickDraw 3D human interface -- Maybe you want to allow users to
visualize 3D data and models in a standalone application or as part of an
existing application. QuickDraw 3D's human interface guidelines and built-in
widgets provide a consistent way of manipulating 3D objects.
- QuickDraw 3D rendering and shading -- Rendering turns the 3D geometries
into pixels; shading determines what color those pixels should be. Realism can
be added by applying textures to objects: texture mapping takes a texture
(usually from a picture source, such as a picture of a brick wall) and wraps it
around an object. For example, Figure 3 shows a dinosaur mesh rendered with a
skin texture picture as a texture map. In its second major release, QuickDraw
3D will enable you to write plug-in renderers and shaders and license them to
other developers.
Figure 3. Dinosaur mesh mapped with a skin-like texture
The dinosaur model was supplied in QuickDraw 3D metafile format courtesy of
Viewpoint DataLabs Intl.
- QuickDraw 3D metafile format -- If you want to provide 3D clip art in the
form of models, you'll really be pleased with QuickDraw 3D's metafile format.
One of the common problems encountered by users when working with several 3D
applications is that of data interchange, where one application's file is not
readable by another due to the multitude of 3D data formats. QuickDraw 3D
addresses this problem by providing a standard for the interchange of 3D data.
This device- and platform-independent representation of 3D data is extensible,
so your custom data gets preserved. And all of the elements for a scene can be
stored in the metafile, including lighting, camera objects, texture maps, and
shaders.
ROAD MAP FOR ADOPTION
Based on our experience working with developers, we've created a road map for
adoption of QuickDraw 3D. Here we'll look at how different application
developers might begin to adopt QuickDraw 3D, in order from the least to the
greatest amount of support. These categories provide you with a general
strategy for bringing QuickDraw 3D into your applications.
- Developers of general-purpose 2D applications should add support for the
metafile format, enabling users to read and save 3D data within an application.
This can be achieved by using the 3D Viewer, which allows 3D objects derived
from metafile data to be viewed and manipulated by the user.
- Developers who use other 3D libraries and may not be ready to move to
QuickDraw 3D just yet should at least add support for the metafile format and
additionally consider adopting the QuickDraw 3D human interface guidelines.
Obviously, support for the metafile format requires writing a parser to convert
metafile data to another internal representation (Apple will be supplying
parser code). Implementing the human interface guidelines will make the
application be compatible with and look consistent with other 3D applications
available on the Macintosh. Note that an application that uses a 3D library
other than QuickDraw 3D will have a harder time using the 3D Viewer.
- Developers of existing 3D applications who want to take the first step
toward creating a QuickDraw 3D-savvy application should take advantage of
QuickDraw 3D's rendering capabilities through the use of immediate-mode
rendering (more on this later). This method provides not only fast rendering in
software but also transparent access to hardware, while allowing the
application to preserve its own data structures. In addition, these developers
should plan to add support for the metafile format and the human interface
guidelines.
- Developers who want to leave the low-level work to QuickDraw 3D, and
concentrate on creating differentiating features within their applications,
should make their applications as QuickDraw 3D-savvy as possible. This means
taking advantage of the full API, including QuickDraw 3D's data structures and
geometries (which provide metafile support virtually for free), rendering (both
immediate and retained modes), and the human interface guidelines.
QUICKDRAW 3D ARCHITECTURE
The QuickDraw 3D architecture isolates in a layer within the system software
those things that all developers have to do, leaving them to concentrate on the
code that will allow their application to stand out. This architecture can be
thought of as a sandwich filling that sits between your application and the
hardware it's running on, isolating you from having to deal with operating
system and hardware issues directly. Like any good sandwich filling, if you
examine it closely, you'll see that it's divided into a number of appetizing
chunks. Figure 4 shows some of the functional blocks that make up QuickDraw 3D,
with an emphasis on those areas that can be customized by developers.
Figure 4. QuickDraw 3D architecture
Let's take a quick look at each of these functional areas, which we'll expand
on later. Here we'll use the word scene to describe not only the objects being
modeled, but also the lighting, camera settings, shaders, and other entities
that affect the final appearance on output devices.
Widgets are used to enhance the user experience for 3D applications. For
example, to allow the user to interact with an object, the application can draw
grab handles, in the form of a translation widget, to allow the object to be
manipulated.
Geometries are the encapsulation of data used to describe an object. Some
geometries are provided as part of QuickDraw 3D, resulting in a very concise
representation; for more information, see "QuickDraw 3D Geometries." (QuickDraw
3D uses geometries to draw widgets.)
QUICKDRAW 3D GEOMETRIES
The QuickDraw 3D geometries that are currently available are as follows: line,
polyline, triangle, point, simple polygon, general polygon, trigrid, mesh, box,
marker, NURB curve, and patch.
In addition, the following geometries are planned for the second major release
of QuickDraw 3D: torus, ellipse, ellipsoid, disk, cylinder, cone, and triangle
strip. (In version 1.0, you can create any of these geometries by representing
them as meshes.)
Where applicable, the geometries are parameterized so that they're ready for
texture mapping or other shading effects.
The
I/O layer provides support for metafiles. There are routines for reading
and writing 3D data to Storage objects, which may be disk or memory based and
are useful for providing Clipboard or drag and drop support in your
application.
Picking is used to determine which object a user chose. QuickDraw 3D's picking
facilities are more extensive than in other 3D libraries, not only providing
several different types but also returning quite a bit of information to the
application beyond whether a hit took place.
Light objects supply the lighting for a scene. QuickDraw 3D provides four types
of light sources: ambient, directional, point, and spot. Based on the light
sources for a given scene and the illumination shader, the renderer makes
intensity calculations for each object's surface and vertex contained in the
scene.
Camera objects define a point of view into a particular scene. QuickDraw 3D
provides three different camera types: view angle, orthographic, and view
plane.
Attributes are used to specify different characteristics for each object (or
parts of an object, such as its vertices or faces), and also to attach custom
data to an object.
Shaders are used to modify or add data, on either a per vertex or a per pixel
basis, as geometries are being processed by the renderer -- for example,
illumination and texturing shaders.
Renderers are the business end of QuickDraw 3D. A renderer is a set of routines
used to create a shaded synthetic model of the scene, based on the information
stored in the geometry and taking into account the lighting, surface
attributes, and camera location. QuickDraw 3D provides two basic renderers: a
wireframe and an interactive renderer. You can extend QuickDraw 3D by
writing a plug-in renderer, developing an accelerator card, or implementing a
combination of both -- a renderer tied to a particular hardware setup.
IMPLEMENTING SUPPORT FOR THE 3D VIEWER
Now, on to the coding details. We realized that some application developers
wouldn't want to get involved with the low-level details of a new API. We
looked at the QuickTime model and saw that a lot of developers implemented
support for viewing movie data by using movie controllers in their existing
nonmultimedia applications. We likewise wanted to allow applications to support
the viewing of QuickDraw 3D metafiles with minimal effort, so we've provided an
additional shared library that implements a 3D Viewer. The Viewer allows users
to view and have a basic level of interaction with 3D data without your having
to make any QuickDraw 3D calls. Figure 5 shows a Viewer implementation in a
modified version of the Scrapbook. (We used a preliminary version, so the
Viewer interface may change.)
Figure 5. Viewer implementation in the Scrapbook
The car model was supplied in QuickDraw 3D metafile format courtesy of
Viewpoint DataLabs Intl.
Adding Viewer support is simple -- it requires only about five function calls.
Your application can check to see if the Viewer is available by calling Gestalt
with the constant gestaltQuickDraw3DViewer.
We'll now look at how your application can create and use a QuickDraw 3D Viewer
object. In the application named Simple 3D Viewer on this issue's CD, we create
a window in which the only object is a Viewer.
As you read through the code samples, you'll notice that function names have a
"Q3" prefix, data types have a "TQ3" prefix, and constants have a "kQ3" prefix.
The part of a function name before the underscore indicates the object being
operated on (the class), while the part after the underscore indicates the
operation (the method). For example, to set the origin of a Box object, you'd
call the function Q3Box_SetOrigin.
CREATING AND DISPOSING OF A VIEWER OBJECT
Creating and disposing of a Viewer object is very easy to do. You attach a
Viewer to a window with the Q3ViewerNew function:
viewerObj = Q3ViewerNew((CGrafPtr)theWindow, &theRect, 0L);
This
function takes a WindowPtr, a pointer to a Rect that describes the window area
where you want the 3D scene to appear, and a long word containing flags for
modifying the behavior of the Viewer. When you're finished with the Viewer, you
need to dispose of it with the Q3ViewerDispose function:
Q3ViewerDispose(viewerObj);
ATTACHING DATA TO THE VIEWER
To display the contents of a metafile in your Viewer, you can use the
Q3ViewerUseFile function:
Q3ViewerUseFile(viewerObj, fileRefNum);
Q3ViewerUseFile
takes a reference to the Viewer object and a file reference to a previously
opened QuickDraw 3D metafile. You can also display data from the Clipboard or
data you created yourself, with the Q3ViewerUseData function:
Q3ViewerUseData(viewerObj, myDataPtr, myDataSize);
This
function takes a reference to a Viewer object, a pointer to the data, and the
size of the data in bytes. The data must be in metafile format.
HANDLING EVENTS
You need to modify your event loop slightly to give the Viewer the opportunity
to handle events, as follows:
wasViewerEvent = Q3ViewerEvent(viewerObj, theEvent);
Q3ViewerEvent
takes a reference to a Viewer object and a pointer to an event record (usually
obtained from WaitNextEvent). This function allows the Viewer to respond to
events, such as a mouse-down event in one of its controls. It returns a value
of type Boolean that indicates whether the event was handled.
If the area occupied by the Viewer needs to be updated, you need to redraw the
data in your update event handler by calling Q3ViewerDraw:
theErr = Q3ViewerDraw(viewerObj);
OTHER VIEWER FUNCTIONALITY
The Viewer allows access to the View object for the scene, which enables you to
customize the Viewer's behavior by changing the renderer or lighting for the
scene (more on Views later). Also, the Viewer provides support for cut, copy,
and paste; see the Simple 3D Viewer sample on the CD for an example.
PROGRAMMING WITH THE QUICKDRAW 3D API: ERROR CHECKING AND INITIALIZATION
Now let's look at programming with the QuickDraw 3D API, starting with error
checking and initialization. First, the QuickDraw 3D shared library needs to be
installed in the Extensions folder or in the same folder as your project.
During your development cycle you should use the debugging version of the
library for extensive error checking.
Error checking may seem like a weird place to start, but checking and
responding to what QuickDraw 3D is trying to tell you will save a great deal of
trouble and strife during development. The QuickDraw 3D error manager provides
several levels of error checking along with functions for checking the last
error that occurred. The error checking, which is similar to that in QuickDraw
GX, has three levels: errors, warnings, and notices.
You can install your own error, warning, and notice
handlers, which can write the error information to a window or file or present
a dialog or alert. Presenting too many alerts can be annoying to the user, so
you should probably log errors, warnings, and notices to a file or a status
window, and present a dialog or an alert only for fatal errors from which no
recovery is possible.
DEFINING AND INSTALLING AN ERROR HANDLER
Handlers for errors, warnings, and notices are all similar -- they're functions
that take an error code of type TQ3Error and have no return value. Listing 1
shows a definition of an error handler.
Listing 1. Error handler
static void MyErrorHandler(TQ3Error firstError, TQ3Error lastError,
long refCon)
{
char buf[512];
sprintf(buf, "ERROR %d - %s\n", lastError,
getErrorString(lastError)); // Get the error as a C string.
if (gErrorFile == NULL)
gErrorFile = fopen("error.output", "w+");
if (gErrorFile != NULL)
fputs(buf, gErrorFile);
}
Once
handlers have been defined, it's a snap to install them. For example, you would
install the error handler defined in Listing 1 as follows:
Q3Error_Register(MyErrorHandler, 0L);
INITIALIZING QUICKDRAW 3D
Before you can use QuickDraw 3D, you need to call Gestalt to see if the library
is installed, using the constant gestaltQuickDraw3D. You then need to
initialize the library as shown in Listing 2. You call the Q3Initialize
function to ensure that the QuickDraw 3D library gets a chance to allocate its
internal data structures and to initialize any subcomponents (such as plug-in
shaders) that it needs to call. You then do other initialization as needed,
such as installing an error handler. The return value indicates whether the
call was successful.
Listing 2. Initializing and closing the connection to the library
void Initialize3DStuff(void)
{
if (Q3Initialize() == kQ3Failure) {
// Handle the error.
StopAlert(kQD3DInitFailed);
ExitToShell();
}
MyErrorInit();
}
void Exit3DStuff(void)
{
if (Q3Exit() == kQ3Failure) {
// Handle the error.
StopAlert(kQD3DExitFailed);
ExitToShell();
}
}
When
your application is about to quit, you should shut down your connection to the
QuickDraw 3D library by calling Q3Exit, also shown in Listing 2. (Obviously a
real application would have more sophisticated error handling here.)
CREATING AND DRAWING A SIMPLE 3D OBJECT: THE BOX APPLICATION
The Box application on this issue's CD is a simple QuickDraw 3D program that
opens a window, displays 3D boxes in the window, and rotates the boxes (see
Figure 6). While this isn't a useful application as such, it covers all
the basics needed to create and display objects using QuickDraw 3D. It also
illustrates double buffering support, which helps an application provide
flicker-free drawing when animating geometries; QuickDraw 3D's double buffering
takes advantage of hardware double buffering when available.
Figure 6. A window from the Box sample program
For a more complex example, see the Modeller program on the CD, which shows
most of the things a QuickDraw 3D application needs to do, such as reading and
writing metafiles, texture mapping, and using interpolation styles.
We define the following data structure to store the information that QuickDraw
3D needs to model and render our scene:
struct _documentRecord {
TQ3ViewObject fView; // The view for the scene
TQ3GroupObject fModel; // Object in scene being modeled
TQ3StyleObject fInterpolation; // Style used when rendering
TQ3StyleObject fBackFacing; // Whether to draw shapes that
// face away from the camera
TQ3StyleObject fFillStyle; // Drawn as solid filled objects
// or decomposed to components
TQ3Matrix4x4 fRotation; // The transform for the model
};
typedef struct _documentRecord DocumentRec, *DocumentPtr,
**DocumentHdl;
We
can create a new instance of this type, initialize it with the required values,
and store a reference to it in each window's refCon field.
OBJECT CREATION
Creating a simple object -- like a box -- is straightforward. We'll make four
copies of the box, each with its own transform. The code to create these boxes
is shown in Listing 3. We can store the boxes in our document simply by storing
the value returned by this function in our document's fModel field.
Listing 3. Creating four boxes
TQ3GroupObject MyNewModel()
{
TQ3GroupObject myGroup;
TQ3GeometryObject myBox;
TQ3BoxData myBoxData;
TQ3GroupPosition myGroupPosition;
TQ3ShaderObject myIlluminationShader;
TQ3Vector3D translation;
TQ3SetObject faces[6];
short face;
// Create a group for the complete model.
if ((myGroup = Q3DisplayGroup_New()) != NULL) {
// Define a shading type for the group and add the shader to
// the group.
myIlluminationShader = Q3PhongIllumination_New();
Q3Group_AddObject(myGroup, myIlluminationShader);
// Set up the colored faces for the box data.
myBoxData.faceAttributeSet = faces;
myBoxData.boxAttributeSet = nil;
MyColorBoxFaces(&myBoxData);
// Create the box itself.
Q3Point3D_Set(&myBoxData.origin, 0, 0, 0)
Q3Vector3D_Set(&myBoxData.orientation, 0, 1, 0);
Q3Vector3D_Set(&myBoxData.majorAxis, 0, 0, 1);
Q3Vector3D_Set(&myBoxData.minorAxis, 1, 0, 0);
myBox = Q3Box_New(&myBoxData);
// Put four references to the box into the group, each one with
// its own translation.
translation.x = 0; translation.y = 0; translation.z = 0;
MyAddTransformedObjectToGroup(myGroup, myBox, &translation);
translation.x = 2; translation.y = 0; translation.z = 0;
MyAddTransformedObjectToGroup(myGroup, myBox, &translation);
translation.x = 0; translation.y = 0; translation.z = -2;
MyAddTransformedObjectToGroup(myGroup, myBox, &translation);
translation.x = -2; translation.y = 0; translation.z = 0;
MyAddTransformedObjectToGroup(myGroup, myBox, &translation);
}
// Dispose of the objects we created here.
if (myIlluminationShader != NULL)
Q3Object_Dispose(myIlluminationShader);
for (face = 0; face < 6; face++) {
if (myBoxData.faceAttributeSet[face] != NULL)
Q3Object_Dispose(myBoxData.faceAttributeSet[face]);
}
if (myBox != NULL)
Q3Object_Dispose(myBox);
return myGroup;
}
Notice
that we dispose of the boxes after adding them to the document group. QuickDraw
3D will create references to the boxes in the document group, so we can safely
dispose of them. To be good QuickDraw 3D citizens and to make more effective
use of memory, we need to dispose of each QuickDraw 3D object as soon as we're
done with it. QuickDraw 3D keeps track of the reference count of each object to
help detect memory leaks. If you're using the debugging version of QuickDraw
3D, it will tell you when you call Q3Exit if there are any objects remaining
that need to be disposed of.
RETAINED AND IMMEDIATE MODE RENDERING
We talked earlier about retained and immediate modes. Which mode to use is the
subject of big philosophical arguments in the world of 3D graphics. Some
developers prefer one over the other as a matter of principle; other developers
make a choice based on the type of application being developed. QuickDraw 3D
offers the best of both worlds: not only does it support both ways of rendering
geometric data, it also allows you to mix these types in the same rendering
loop.
Retained and immediate modes are simply methods of rendering, without the usual
connotation of the term "mode" (a state that you must exclusively remain in
once you get into it). Although this terminology has become common in the field
of 3D graphics, retained rendering and immediate rendering calls can in fact be
freely mixed.
In
retained mode, the definition and storage of the geometries are kept
internal to QuickDraw 3D (as an abstract object). This mode provides convenient
features for caching, rejection of entire objects based on clipping and
culling, preservation of tessellated surfaces, multiple instantiation of
objects (drawing multiple versions of an object but storing its definition only
once), and conversion between geometry types. Retained mode is useful when the
geometry has to be passed around to different modules within the application or
to plug-in renderers. Extensive geometry editing functions are provided as part
of the QuickDraw 3D API, which makes it easy to alter the data associated with
each geometric object.
In immediate mode, the application keeps the only copy of the geometry. This is
particularly useful when your application needs to reference data that's in a
format different from the one used by QuickDraw 3D or when a large number of
vertices that make up the geometry are being edited continuously -- for
example, in the animation of a stress analysis for mechanical design.
The code in Listing 3 creates the boxes in retained mode, by creating objects
that encapsulate the box data; QuickDraw 3D then manages the box data for
us. If you want to add QuickDraw 3D rendering and drawing to an existing
application with its own 3D data structures, you can draw in immediate mode
instead. To draw a box in immediate mode, you simply initialize the values in
the TQ3BoxData structure to the appropriate values and then draw the data
directly in a rendering loop (described later) by calling the following
function:
myStatus = Q3Box_Submit(&myBoxData);
Because
you never create a QuickDraw 3D object, there's no need to call
Q3Object_Dispose.
Notice that in Listing 3 we initialize an object using a data structure of type
TQ3BoxData. This structure contains all of the information required to draw a
Box geometry, but is not an object in itself. Because of this we don't call
Q3Object_Dispose on the box data structure, but we do call it on the Box
object.
THE DRAW CONTEXT
All window system dependencies are isolated to a layer we call the draw
context. This makes porting your application easier (and it also makes it
easier for us to port QuickDraw 3D to other platforms). Although QuickDraw 3D
is platform independent, of course at some point you'll need to deal with the
realities of a particular platform's window system, in this case the Mac OS.
This is where the concept of a draw context comes in. It's a means for
QuickDraw 3D to interface with the host environment. There's a special draw
context for the Mac OS, called a Macintosh draw context; information describing
this context is stored in a TQ3MacDrawContext object, which contains the
information necessary for QuickDraw 3D to image the data on a computer running
the Mac OS.
Listing 4 is a routine from the Box application that creates a Macintosh draw
context the size of a window that we pass in. We're telling QuickDraw 3D to
create a buffer in which to image the data; this is referred to as the back
buffer. If we're using double buffering (that is, we set the doubleBufferState
field of the Macintosh draw context to true), the front buffer will be the
window associated with the draw context. The data is copied from the back
buffer to the front buffer when Q3View_EndRendering is called. This helps
provide flicker-free animation if you're animating the object being viewed.
Listing 4. Creating a Macintosh draw context
TQ3DrawContextObject MyNewDrawContext(WindowPtr theWindow)
{
TQ3DrawContextData myDrawContextData;
TQ3MacDrawContextData myMacDrawContextData;
TQ3DrawContextObject myDrawContext;
TQ3ColorRGB clearColor;
Q3ColorRGB_Set(&clearColor, 1, 1, 1);
myDrawContextData.clearImageState = kQ3True;
myDrawContextData.clearImageMethod = kQ3ClearMethodWithColor;
myDrawContextData.clearImageColor = clearColor;
myDrawContextData.paneState = kQ3False;
myDrawContextData.maskState = kQ3False;
myDrawContextData.doubleBufferState = kQ3True;
myMacDrawContextData.drawContextData = myDrawContextData;
myMacDrawContextData.window = (CGrafPtr) theWindow;
// The window associated with the view
myMacDrawContextData.library = kQ3Mac2DLibraryNone;
myMacDrawContextData.viewPort = nil;
myMacDrawContextData.grafPort = nil;
// Create draw context and return it; if nil, caller must handle
// it.
myDrawContext = Q3MacDrawContext_New(&myMacDrawContextData);
return myDrawContext;
}
Sometimes
you might want to be able to get at the back buffer yourself; for example, you
might want to create a picture preview of some metafile data to place on the
Clipboard along with the metafile data, so that applications that don't support
metafiles can display the picture. QuickDraw 3D makes this possible by
providing a different type of draw context, called a pixmap draw context, which
can be based on a GWorld. First you need to create a GWorld the size of the
window area; then you can create a pixmap draw context as shown in Listing 5.
Listing 5. Creating a pixmap draw context
TQ3DrawContextObject MyNewPixmapDrawContext(GWorldPtr theGWorld)
{
TQ3PixmapDrawContextData myPixmapDCData;
TQ3ColorRGB clearColor;
PixMapHandle hPixMap;
Rect srcRect;
Q3ColorRGB_Set(&clearColor, 1, 1, 1);
// Fill in the draw context data.
myPixmapDCData.drawContextData.clearImageState = kQ3True;
myPixmapDCData.drawContextData.clearImageMethod =
kQ3ClearMethodWithColor;
myPixmapDCData.drawContextData.clearImageColor = clearColor;
myPixmapDCData.drawContextData.paneState = kQ3False;
myPixmapDCData.drawContextData.maskState = kQ3False;
myPixmapDCData.drawContextData.doubleBufferState = kQ3False;
hPixMap = GetGWorldPixMap(theGWorld);
LockPixels(hPixMap);
srcRect = theGWorld->portRect;
myPixmapDCData.pixmap.width = srcRect.right - srcRect.left;
myPixmapDCData.pixmap.height = srcRect.bottom - srcRect.top;
myPixmapDCData.pixmap.rowBytes = (**hPixMap).rowBytes & 0x7FFF;
myPixmapDCData.pixmap.pixelType = kQ3PixelTypeRGB32;
myPixmapDCData.pixmap.pixelSize = 32;
myPixmapDCData.pixmap.bitOrder = kQ3EndianBig;
myPixmapDCData.pixmap.byteOrder = kQ3EndianBig;
myPixmapDCData.pixmap.image = (**hPixMap).baseAddr;
return Q3PixmapDrawContext_New(&myPixmapDCData);
}
When
using a pixmap draw context, you must keep the GWorld's PixMap locked all the
time (which implies that you need to call LockPixels on it, to help avoid heap
fragmentation). Also, the PixMap must be 32 bits deep -- other depths are not
supported.
THE CAMERA
A camera is a QuickDraw 3D object used to project a 3D scene onto a 2D plane.
It defines a point of view on the scene and a method of projection onto the
viewing plane. QuickDraw 3D provides three types of camera:
- View angle or perspective -- This type of camera is defined in terms of a
viewing angle and horizontal-to-vertical aspect ratio. It's the most common
camera type because it provides a natural-looking perspective.
- Orthographic -- This is a parallel projection, where the direction of
projection is perpendicular to the projection plane. Orthographic projections
are generally less realistic than perspective projections; however, they're
popular for engineering drawings because parallel lines remain parallel in the
projection, rather than converging to a single point on the horizon.
- View plane -- This is a perspective projection defined in terms of an
arbitrary viewing plane. This type of camera is useful for providing an
off-axis view, which is used for scrolling.
We use a view angle camera for
the Box application, creating the camera with the routine in Listing 6.
Listing 6. Creating the camera
TQ3CameraObject MyNewCamera(WindowPtr theWindow)
{
TQ3ViewAngleAspectCameraData perspectiveData;
TQ3CameraObject camera;
TQ3Point3D from = { 0.0, 0.0, 13.0 };
TQ3Point3D to = { 0.5, 0.5, -1.5 };
TQ3Vector3D up = { 0.0, 1.0, 0.0 };
float fieldOfView = 0.523593333;
float hither = 0.001;
float yon = 1000;
perspectiveData.cameraData.placement.cameraLocation = from;
perspectiveData.cameraData.placement.pointOfInterest = to;
perspectiveData.cameraData.placement.upVector = up;
perspectiveData.cameraData.range.hither = hither;
perspectiveData.cameraData.range.yon = yon;
perspectiveData.cameraData.viewPort.origin.x = -1.0;
perspectiveData.cameraData.viewPort.origin.y = 1.0;
perspectiveData.cameraData.viewPort.width = 2.0;
perspectiveData.cameraData.viewPort.height = 2.0;
perspectiveData.fov = fieldOfView;
perspectiveData.aspectRatioXToY =
(float)(theWindow->portRect.right - theWindow->portRect.left) /
(float)(theWindow->portRect.bottom - theWindow->portRect.top);
camera = Q3ViewAngleAspectCamera_New(&perspectiveData);
return camera;
}
LIGHTING
QuickDraw 3D includes a number of different light objects that can be used to
provide illumination to the surfaces in a scene. Lighting is additive, meaning
that the amount of lighting applied to a particular surface will be the sum of
the lighting from all sources. There are four light types:
- Ambient -- This is the amount of light added to all surfaces in a scene.
Since this light type has no location, it doesn't cast shadows.
- Directional -- Sometimes referred to as an "infinite" light, this light
source emits parallel rays of light in a specific direction. The intensity of
this light source doesn't change as the distance from the light changes.
- Point -- This light source emits rays of light in all directions from a
particular point location. A point light is attenuated, meaning that the
intensity of the light decreases as the distance from the light increases;
QuickDraw 3D provides a set of constants to control this behavior.
- Spot -- This type of light emits a circular cone of light from a point
source in a particular direction. A spot light is attenuated both by the
distance from the source and by the position across the cone; the intensity of
light at the center of the cone is greater than the intensity at the edge of
the cone.
Listing 7 shows an extract from our sample's MyNewLights routine;
here we create a point light and add it to a light group.
Listing 7. Creating a point light in a light group
lightGroup = Q3LightGroup_New();
pointData.lightData.isOn = kQ3True;
pointData.lightData.brightness = 0.80;
pointData.lightData.color.r = 1.0;
pointData.lightData.color.g = 1.0;
pointData.lightData.color.b = 1.0;
pointData.location.x = -10.0;
pointData.location.y = 0.0;
pointData.location.z = 10.0;
pointData.castsShadows = kQ3False;
pointData.attenuation = kQ3AttenuationTypeNone;
light = Q3PointLight_New(&pointData);
Q3Group_AddObject(lightGroup, light);
Q3Object_Dispose(light);
THE VIEW
Once you've added the light to a group, you can associate the group with the
View object for your scene. A View object keeps track of the information
necessary to render an entire scene, tying together the different parts of
QuickDraw 3D. In our simple example it ties together the draw context, camera,
lights, and renderer. Listing 8 shows the code we use to create the View object
for the Box program.
Listing 8. Creating the View object
TQ3ViewObject MyNewView(WindowPtr theWindow)
{
TQ3Status myStatus;
TQ3ViewObject myView;
TQ3DrawContextObject myDrawContext;
TQ3RendererObject myRenderer;
TQ3CameraObject myCamera;
TQ3GroupObject myLights;
myView = Q3View_New();
// Create and set the draw context.
myDrawContext = MyNewDrawContext(theWindow);
myStatus = Q3View_SetDrawContext(myView, myDrawContext);
Q3Object_Dispose(myDrawContext);
// Create and set the renderer. Use the interactive software
// renderer.
myRenderer = Q3Renderer_NewFromType(kQ3RendererTypeInteractive);
myStatus = Q3View_SetRenderer(myView, myRenderer);
Q3Object_Dispose(myRenderer);
// Create and set the camera.
myCamera = MyNewCamera(theWindow);
myStatus = Q3View_SetCamera(myView, myCamera);
Q3Object_Dispose(myCamera);
// Create and set the lights.
myLights = MyNewLights();
myStatus = Q3View_SetLightGroup(myView, myLights);
Q3Object_Dispose(myLights);
return myView;
}
THE RENDERING LOOP
All drawing must be done in a rendering loop. This is necessary because we
don't know in advance how much memory is required to render a complex model.
The loop should fit neatly into your application, because most Macintosh
applications will localize drawing in the update event-handling code, which is
where you'll call your rendering loop for QuickDraw 3D. A simple rendering loop
will look like Listing 9.
Listing 9. The rendering loop
TQ3Status DocumentDraw3DData(DocumentPtr theDocument)
{
Q3View_StartRendering(theDocument->fView);
do {
Q3Style_Submit(theDocument->fInterpolation,
theDocument->fView);
Q3Style_Submit(theDocument->fBackFacing, theDocument->fView);
Q3Style_Submit(theDocument->fFillStyle, theDocument->fView);
Q3MatrixTransform_Submit(&theDocument->fRotation,
theDocument->fView);
Q3DisplayGroup_Submit(theDocument->fModel, theDocument->fView);
} while (Q3View_EndRendering(theDocument->fView)
== kQ3ViewStatusRetraverse);
return kQ3Success;
}
Recall
that earlier we set up our Macintosh draw context to use double buffering; this
causes all drawing to take place in the back buffer. The calls in the rendering
loop draw into the active buffer, which we have set up to be the back buffer.
The image data is copied from the back buffer to the front buffer (in this case
the window) when Q3View_EndRendering is called.
A rendering loop for a pixmap draw context would be similar to the routine in
Listing 9, except you would need to copy the data from your PixMap to the
screen yourself, generally with CopyBits.
THE QUICKDRAW 3D METAFILE
Here we'll take a brief look at the architecture of QuickDraw 3D's metafile
format (file type '3DMF') and at how you can provide metafile support in your
application.
The QuickDraw 3D metafile comes in two forms: plain-text (ASCII) and binary.
Table 1 shows the differences between these two forms. The plain-text form is
more useful for debugging purposes; once your application is debugged, it's
more efficient to use the binary form, which may be read and written much
faster and may require less storage space on disk.
The metafile format supports a wide range of primitive data types, including
1-, 2-, 4-, and 8-byte signed and unsigned integers and 4- and 8-byte IEEE
floating-point numbers, together with a range of types for describing 3D data.
In addition, metafiles may contain big- or little-endian numbers, making them
ideal for storing data that may be used in a cross-platform manner.
METAFILE ORGANIZATION
There are three distinct types of metafile organization: normal, stream, and
database. The organization of the file can affect both the size of the file and
the time it takes to access the data in the file. Let's look at a simple
example in which a single Box object is drawn four times at different positions
by means of four different Transform objects, as was shown in Figure 6. The
three types of organization are illustrated in Figure 7. (Note that # marks the
beginning of a comment.) These types are as follows:
- Normal -- This is the most compact representation. Referenced objects are
listed in a Table of Contents (TOC). In our example, only the Box object is
listed in the TOC. The Transform objects don't appear in the TOC because they
were referenced only once. Note that random access to the file is needed to
resolve references, since after reading a reference, the metafile parser needs
to skip forward to the TOC, and back to resolve the references.
- Stream -- There is no TOC, and references to objects are written as
copies of the objects themselves. This may result in a larger file if a lot of
object references were used, but it allows for a sequential search. A
sequential search is very useful for reading from the file and imaging to a
printer, since each object can be read, imaged, and disposed of. This
organization is also useful as a wire protocol for imaging on remote machines.
- Database -- Every object is logged into the TOC, even if it's not
referenced. Each TOC entry contains the type of the object. Accessing the TOC
lets you see all the information contained in the file without having to read
in all of the file and create objects. This would be useful for creating a
catalog of textures, for example.
Figure 7. Three types of metafile organization (representing Figure 6)
USING METAFILES
The simplest way to access data in metafiles is to use the QuickDraw 3D API.
First, there are two types of objects you need to understand:
- TQ3FileObject -- Objects of this type maintain state information and
provide an interface between a given file format and a Storage object. File
objects are used to read and write data in metafile format from and to Storage
objects.
- TQ3StorageObject -- Objects of this type act as an interface to a type
of physical stream-based storage (for example, memory and files). Storage
objects are used to represent a piece of physical storage.
Why have this
two-stage approach? The answer is that all the machine dependencies are
localized in the Storage object, which allows files to be used to read and
write data from differing types of physical storage with the same set of
routines. For example, you can use the same File object to write to a Storage
object that represents a file on your hard disk and to write to another Storage
object that represents a block of memory that will be passed to the Scrap
Manager.
The usual method for using File and Storage objects is to create a new instance
of a Storage object and attach it to a newly created File object using
Q3File_SetStorage, as shown in Listing 10.
Listing 10. Attaching a Storage object to a file
TQ3FileObject MyGetNewFile(FSSpec *myFSSpec, TQ3Boolean *isText)
{
TQ3FileObject myFileObj;
TQ3StorageObject myStorageObj;
OSType myFileType;
FInfo fndrInfo;
// We assume the FSSpec passed in was valid and get the file
// information. We need to know the file type; this routine may
// get called by an Apple-event handler, so we can't assume a
// type -- we need to get it from the FSSpec.
FSpGetFInfo(myFSSpec, &fndrInfo);
myFileType = fndrInfo.fdType;
if (myFileType == '3DMF')
*isText = kQ3False;
else if (myFileType == 'TEXT')
*isText = kQ3True;
else
return NULL;
// Create a new Storage object and new File object.
if (((myStorageObj = Q3FSSpecStorage_New(myFSSpec)) == NULL)
|| ((myFileObj = Q3File_New()) == NULL)) {
if (myStorageObj != NULL)
Q3Object_Dispose(myStorageObj);
return NULL;
}
// Set the storage for the File object.
Q3File_SetStorage(myFileObj, myStorageObj);
Q3Object_Dispose(myStorageObj);
return myFileObj;
}
Reading data from metafiles. There are three routines that you can use to help
with reading the data: Q3File_GetNextObjectType, Q3File_ReadObject, and
Q3File_SkipObject. Listing11 illustrates the technique used to read
drawable data from a metafile. The code loops through the file, getting each
object and checking to see if the object is drawable; if so, it adds the object
to a group object.
Listing 11. Reading from a metafile
TQ3Status MyReadModelFromFile(TQ3FileObject theFile, TQ3GroupObject
myGroup)
{
if (theFile != NULL) {
TQ3Object myTempObj;
TQ3Boolean isEOF;
// Read objects from the file.
do {
Q3File_ReadObject(theFile, &myTempObj);
if (myTempObj != NULL) {
// We want the object in our main group only if we can
// draw it.
if (Q3Object_IsDrawable(myTempObj))
Q3Group_AddObject(myGroup, myTempObj);
// We either added the object to the main group, or we
// don't care, so we can safely dispose of it.
Q3Object_Dispose(myTempObj);
}
// Check to see if we've reached the end of the file yet.
Q3File_IsEndOfFile(theFile, &isEOF);
} while (isEOF == kQ3False);
}
if (myGroup != NULL)
return kQ3Success;
else
return kQ3Failure;
}
Because
we're isolating the implementation details of how the metafile data is stored
in the Storage object that we associated with the File object at its creation
time, we don't care how the metafile data we're reading is physically stored.
What this means is that we could use the routine above to read data from the
scrap, from a handle supplied by the Drag Manager, or from a file, as long as
the storage object attached to the file is set up properly.
Writing data to metafiles. Data is written to files similarly to the way it's
drawn in a rendering loop. Depending on the available memory and the complexity
of the model, QuickDraw 3D may need to traverse the model in the group more
than once in order to write all the data out (this is the same reason that the
rendering needs to be done in a loop). As shown below, you need to preface your
file-writing loop with a call to Q3File_BeginWrite, and test the value returned
by Q3File_EndWrite to see if there's a need to traverse the data again.
Q3File_OpenWrite(file, kQ3FileModeNormal);
Q3File_BeginWrite(file);
do {
Q3Object_Write(group, file);
} while (Q3File_EndWrite(file) == kQ3FileStatusRetraverse);
Q3File_Close(file);
GO TO IT!
QuickDraw 3D lowers the bar for application developers who want to put support
for 3D data into their applications. By providing support for the features that
all developers need to have in applications -- geometries, metafile support,
rendering, and human interface -- QuickDraw 3D allows you to concentrate on the
features and facilities that set your application apart.
PABLO FERNICOLA (AppleLink PFF, eWorld EscherDude) After spending many years
working in 3D graphics under operating systems named **IX, in a faraway land
called Alabama, Pablo made the transition to real computers. After moving to
Silicon Valley, he learned to beat the traffic jams by getting to work before 8
A.M. and going home after 10 P.M. Now he can be found staring out the window
and wondering how he's going to get home on Interstate 280 after the next
earthquake.
NICK THOMPSON (AppleLink NICKT) is currently establishing himself as the
Mountain Dew-guzzling fat fool of Developer Technical Support. Unable to work
the winter blubber off due to killer waves that are preventing him from surfing
on the California coast, Nick has been consoling himself with learning the
wonder that is QuickDraw 3D. He was last seen wandering down one of the
corridors at Apple mumbling to himself.
Thanks to our technical reviewers Kent Davidson, Eiichiro Mikami, Don Moccia,
and Dan Venolia, and to all the members of the QuickDraw 3D team. Special
thanks to Kent and Dan for supplying information used in this article and to
David Vasquez for his Viewer sample. Thanks also to the Shawn and John team
(Shawn Hopwood, Apple's 3D evangelist, and our marketing weenie, John Alfano)
for their input.
