The Real-Time SpanArray-based Z-buffered 3D Renderer

FEATURING:

• polygon-based model rendering includes convex, concave, and self-intersecting polygons
• fully-generalized hidden surface removal allowing for protruding surfaces based on a SpanArray that is a hybrid between a full Z-buffer and a scanline algorithm for determining visible surfaces
• full-texture mapping with tiling and transparency
• full 3D clipping to the viewplane and the sides of the viewplane window
• movable front clipping plane allowing a cut-away view of the "inside" of models
• independent object and world coordinate system transformations
• complete hierachical modeling system (for designing motion object-dependent systems such as robot arms, etc.)
• complete 6DOF arbitrary camera viewing system (to accompany world coordinate transformations)
• written entirely in C and C++ and merely writes to a framebuffer so it can easily be ported to other platforms where the contents can be copied to a video display device or to an image file
• simple scene and object definition data files define a "frame"
• fairly complex animation sequences can be rendered quite easily by taking advantage of the movable camera and the object hierarchy.
• light-mapping to allow for incremental shading using only 256 colors. An "optimal" palette is chosen for the desired "ideal" range of intensities.
• backface culling has been added for enclosed-volume objects.
• full Phong-illumination model (ambient, diffuse, and specular components)
• Phong shading (not yet in realtime, but will be)

SAMPLE OUTPUT:

These two early renderings show a sequence of animation showing the texture mapping and object protrusion. Only object transformations were implemented and the camera was fixed facing down the Z-axis.

Interacting scene of objects that have their own transformations (such as spinning about their own axes, etc) along with world transformations that move the entire scene as one object. Notice the transparent texture of the cube on the right.

Same scene after a few seconds of animation. Each of the objects have moved independently of each other and the world transformation has shifted the viewing position.

The same scene with the objects and world in a different orientation.

Three Phong-shaded cubes with full Phong illumination model (ambient, diffuse, and specular components). This still uses a 256-color palette that has been optimally fitted using a lightmap table that finds the "closest" intensity to the "ideal" (I devised a routine that can recalculate this table in "near" realtime - in under 2 seconds, but it only needs to be updated when the lighting changes).