Rendering is the process of creating two-dimensional computer graphics images (i.e. pixel arrays) from three-dimensional scene databases.

Projection - like casting a shadow

During the rendering process, a 3-D scene database is projected onto a 2-D virtual image plane defined by a virtual camera.

• Rasterization and aliasing

 

• Improve resolution via:




• Pixels
• width x height
• Colors
• For n bits, number of colors = 2 to the n power

• Geometry Calculations - transforming object space coordinates into screen space coordinates, clipping unseen portions of the database, tessellating the database, determining vertex colors based on surface definitions and light falling on objects

 
• Triangle Setup Calculations - determining the horizontal and vertical rates of change in color (in screen space) between vertices of triangles

 
• Rasterization - generating pixel-based data from the results of the above steps

 
• Frame buffer - display memory into which pixel data is written and from which the display is refreshed (alternatively can render to a disk file)

 

Software, hardware and the rendering pipeline:

• The core application is implemented in software

 
• The frame buffer by definition is implemented in hardware

 
• The intervening steps in the rendering pipeline can be handled in either hardware or software

 
• Increasingly, they are being handled in hardware, which is one of the key forces that drives increasing price/performance in graphics systems

 

Various algorithms then may be employed for converting the 2-D image to a pixel array appropriate for a given display, and determining the colors for each pixel in that array. Pixel colors are derived from the interplay of various elements of a 3-D scene. The color of each pixel in the rendered image potentially is influenced by:

• Position and orientation of scene geometry
• Postion and optical characteristics of the camera

 
• Material and surface definitions

 
• Number, color, type and position of light sources

 
• Environments, Atmospheres, etc.

 
• Rendering algorithm(s) employed

 

What constitutes realism?
 

• the rigorous simulation of the physical phenomena of light bouncing through an environment, in which an attempt is made to mimic the workings of reality, usually more accurate, but slower

 
• the illusion of realism by empirical approximations, in which an attempt is made to represent an impression that is sufficiently similar to reality to serve a particular purpose, usually faster, but ``good enough''

 

Why not always strive for maximum ``realism''?
 

Because higher fidelity feedback usually has a cost. Satisfactory performance may require:
 

• More sophisticated graphics hardware ($)

 
• Longer time to render (decreased interactivity, increased production time)

 

Factors affecting rendering efficiency:

• complexity of geometry (number of entities/polygons)


more geometric detail = more calculations = more time to render
 

• complexity of surface representation (shading models, surface mapping, etc.)


richer surface descriptions = more calculations = more time to render
 

• complexity of lighting


more lights or more complex lights = more calculations = more time to render
 

• complexity of rendering algorithm

 

When do we render?

Constantly, at various levels of fidelity, at various stages in the 3-D process, depending on the needs of the moment.

A pragmatic approach to rendering requires that we determine what level of interactivity will be required and what level of representation is ``good enough'' for a given case, and then tailor the rendering approach to those requirements.

Back to 3-D Color outline

This file was last modified on September 23, 2002.