Computer Graphics

Introduction

Use of computers to generate images for

• Visualization (simple data, medical)
• Industrial design (electrical, mechanical)
• Animation (movies and visualization)
• Games

Requires

• Lots of Math
  
  • Linear Algebra (matrices, vectors: dot products)
  • Discrete representations (images)
• A little Physics
  
  • Lighting/Energy Transfer
  • Color reflections/refraction
• A whole lot of Computer Science
  
  • Data structures (arrays, dynamics structures)
    
  • Languages

How?

• Map world scene to pixels on the screen (Math).
• Pixels are dependent on many physical properties (Physics).
• Scan line
• Ray Tracing

---

OpenGL

General purpose, open source, Graphics code:

• Widely used
• Widely available:
  (bindings for C, C++, FORTRAN, Java, Python, Ruby, Tcl)
• Powerful (lighting and shading built-in)
• Easy to use

Visualization

Typically interested in rendering volumetric data on a 2D surface (screen).
Requires knowledge of:

• Computer Graphics
• Image Processing

A considerable amount of code---Software Engineering is important!

How?

• Voxels (Volume elements)
• Very computationally intensive!

Visualization ToolKit---VTK

• Coding Standards
• VTK in Use
• Visible Human

Advanced Visual Systems

• Toolmaster
• AVS Express

IBM Visualization Data Explorer (DX)

• Image Gallery
• Color Map

MPIRE Massively Parallel Interactive Rendering Environment

• The Splatter Image Gallery
  The images were generated using Splatter on the Cray T3D (128 processors, 8GB of RAM) and Cray T3E (240 processors, 32GB of RAM) at the San Diego Supercomputer Center.
• The Splatter Movie Gallery
  The animations were generated using Splatter on a Cray T3D equipped with 128 processors and 8GB of RAM.
• Splatter Users Manual
  For some perspective, rendering a 512x512x512 data set requires a minimum of 4 bytes per voxel (512MB). The Visible Human Project (VHP) has a CT data set that is 512x512x1873 (1.8GB to render). The VHP also have two cryosection data sets that are 2048x1216x1878 (14 GB) and 2048x1216x5189 (38 GB) in size. To support interactive volume rendering, the entire data set must reside in RAM---waiting for 1.8GB to be read from disk for each image is not an option. The increasing availability of parallel computers with hundreds or thousands of processors provides a viable solution. For instance, the CRAY T3E available at SDSC has 256 processing elements (PEs) with 128MB of RAM per PE (for a total of 32GB).

Conclusion

• Graphics and visualization techniques are used in many different fields:
  Games to Space Exploration
• Computer Science makes it happen!

Created by Bruce Bolden on Tuesday, December 4, 2001.