CS789, Spring 2007

Lecture notes, Week 6


Output Concepts

Quite a lot of our discussion when we considered input concepts centred around the "echo" component of the interaction protocol, which we also often called feedback. Thus, we inevitably talked about output, and using a set of concepts closely allied to the concepts we used when talking about input. Rather than reviewing them explicitly in this lecture we will just let them come up as they occur, while following a quite different organization, centred around the concept of "image quality".

Attention

Attention is the most precious resource a user possesses. It can be evoked by output in several ways

Output that requires no attention can also be provided.


Image Quality

What is "image quality"? The phrase is derived from visual devices, and is an amalgam of a few things:
  1. How close is an image to the desired image?
  2. How much information can the device provide?
  3. How accessible to the user is the information provided?

Examples of Output Devices

The key concepts (not orthogonal) are:
  1. Dimensionality: Where an item of output occurs is limited by physics to three space dimensions and one time dimension, but may interact with more than one modality. But,
  2. Attention.
  3. Integration with input
  4. Soft/hard. This is a concept related to time, and to the duration of the signal. Hard allows a different use of attention, contemplative, than does soft, focussed.
  5. Synchronous/asynchronous. Does the user have to coincide with the output in space or time?
  6. Sampled and reconstructed output. Our output devices are normally discrete, not analogue
  7. Levels of performance are affected by sampling, but also by coding. The more redundant the code, the more forgiveness there is for failures of attention.

We can now expand these concepts across a variety of devices

  1. Visual devices
  2. Auditory devices
  3. Tactile devices
  4. Olfactory devices

Image Quality Concepts

When people talk about the goodness and badness of output devices they seem to focus a lot on something like "fidelity" to some standard. This is the concept I generically call "image quality". Where does this idea come from? And how has it evolved?
  1. Where is came from. The device is the output of an imaging system
  2. Concept breaks when there is no original.
  3. Now we can select from a sequence of concepts
    1. Reproduce as many aspects of the original as possible.
    2. Reject spurious information, by filtering. E.g., anti-aliasing.
    3. Enhance important information,
      • often to overcome short-comings of the output system
      • make visible something that would otherwise be hidden
      • objective? customizable? rules?
    4. Provide whatever information the user needs
      • What information is needed? task requirements, user requirements
      • How is it most easily comprehended?
      • Should extra information be provide? (Just enough information is potentially totalitarian.)
      • Where should information selection occur?
        • in the user?
        • in the interface?

Device Properties that Influence Image Quality

Output devices that reconstruct output from samples have a fairly standard set of properties that affect their peformance. It's easiest to understand them by considering how they greate their output.

At each pixel location the device accepts a weight as input and multiplies it by the pixel profile, which is then added to the output. Then...

  1. There are the pixel profiles themselves
  2. Are the pixels at different locations independent of one another?
  3. What human parameters affect perception of continuity?
  4. How should the device be adjusted

Return to: