Basic Hardware Overview

• If you have book access, this is mostly from chapter 1.
• Biology and physics dictates the colors we see.
  • I am not going into that .
• We can do colors two ways
  • Much of art uses a subtractive color system.
    • You start with a white surface
    • You add pigments that "remove" the light that is reflected.
    • In this system, cyan, yellow and magenta are usually the primary colors.
  • In most computer display systems we use an addative color system.
    • We start with a black screen
    • Then add components of light to increase what is emitted.
    • In this system, red, green and blue are the primary colors.
• What?
  • Originally a monitor was a glass plate coated with phosphor
    • An electron beam would "strike" the plate and the phosphor would "glow" for a short period of time.
    • The beam was controlled by a "gun", which would sweep across the screen
    • 
    • the gun would shoot through a "grid" which would discretize the output.
    • The light would only glow for a short period of time, and needed to be "refreshed", or another pass over the screen.
    • So the image would need to be "stored" in memory.
    • This was called the frame buffer
  • The frame buffer was just some memory to store if a pixel was on or off.
    • The gun would "fire" if the pixel were on, and not fire if the pixel were off.
    • Thus a 600x500 display would need 300,000 bits to store the image.
    • or 36KB (bytes)
    • Since the gun would move from left to right, top to bottom, the tradition became to make the upper left of the screen 0,0
  • At some point they decided to fire the gun at different intensities.
    • Somewhere between 0 and 100%.
    • With the intensity represented by an integer.
    • Usually a power of 2, so the numbers worked nice with memory.
    • An 8 bit gray scale display would display colors between 0 and 255.
    • Remember, only one coat of phosphor
    • But we now need 8 bits per pixel.
    • 600x500x8 = 2,400,000 bits or 292KB.
  • So why not add more phosphor
    • 
    • That way we could fire three different guns and get different colors.
    • But now we need 8x3 bits per pixel.
    • 4.5 MB memory.
  • While modern technology has changed
    • The coordinate system is still here.
    • We still think in terms of pixels.
    • We still think in terms of a frame buffer, on a chunk of memory storing information about how we will "light" the pixels.
      • But we often augment this with additional information.
      • depth, alpha value, ...
• Ok but ...
  • In graphics, we generally view the word through a "synthetic camera"
  • 
  • Light strikes the images in the scene and travels through the "lense" of the camera to strike a "photographic plate" in the back of the camera.
  • The intensity and color of the light determines what is recorded on this photographic plate.
  • We frequently reduce light to be a point source or assume that light from a single source is emitted from a single point in all directions.
    • This is a simplifying assumption.
    • Light sources are usually bigger. (think wire in a light bulb, or the sun)
    • But it makes our life more simple.
• So a model
  • We generally have objects and light sources in a scene
  • Both have properties
    • Common properties like location.
    • Lights have properties like color of light emitted.
    • Objects have properties like the color of light reflected and amount.
  • The viewer, or camera is also important
    • Location, orientation, ...
  • In our model
    • Light from a light source
      • Strikes the viewer
      • Strikes an object
      • Travels off into space.
    • If it strikes the viewer, it is added to the "image"
    • If it travels off into infinity, it is discarded.
    • If it strikes an object it is
      • Absorbed by the object.
      • Reflected off the object with altered properties.
      • Transmitted through the object with altered properties.
    • Absorbed light is ignored.
    • Reflected and transmitted light, see light source.
  • 
  • There is just one problem with this model...
    • There are infinitely may light rays at each source.
    • And this can be multiplied by reflected and refracted light at a given point.
  • A solution is reverse the situation, Ray-casting.
    • 
    • Instead of tracing light from the source, trace it from the destination.
  • A ray can
    • Travel to infinity, pixel is black.
    • Hit a light source, pixel is the color of the light source.
    • Hit an object.
  • When a ray hits an object, we need to compute the color of the object at that point.
    • From the point, cast a ray to each light source.
      • If it hits, add the color of that light source as it interacts with the material at that point.
      • If it hits another object first, recursive case.
    • If the object is transparent or translucent
      • Generate rays passing through the object as well.
  • 
• We will actually end up doing something different from this for quite a while.
  • We will do a "cheap" light computation on the objects.
  • We will then "project" the objects to the photographic plate
  • Then use a line drawing algorithm to turn this into pixels.
  • 
• So back to colors
  • We will use RGB as our primary color system.
  • We are dealing with adding light after all.
  • We will assume that lights have RGB properties.
    • A light turned off has rgb values of (0,0,0)
    • A light with all three values at full has a value of (1,1,1)
    • We might end up scaling these between 0 and N
      • Like 0 to 255 for HTML
  • Look RGB (Red, Green, Blue)