HCI and Usability Issues of Multimedia Internet broadcasting Lecture 3

Preview:

Citation preview

HCI and Usability Issues of Multimedia Internet

broadcasting Lecture 3

Introduction

• Using Multimedia Internet Broadcasting– Appears simple– Controls like CD or VCR player

• But has variable bandwidth - this leads to – loss of service– reduced service parameters

• Needs QoS contract

QoS and usability

• Two sides to the problem (see Lecture 1)

– The tools and methods available to the broadcaster to create the options and

– The tools and methods available to the viewer to evaluate the options

Conflicting Goals

• Broadcasters– To offer options that attract viewers– Increase audience size – Reach all likely users

• Users– To get the highest QoS with a given budget– To maintain service through a broadcast

The Problem

• Choice of package determines– audience and– parameters available for choice

• Packages allow choice of – compression algorithm– audio compression rates– video compression rates– video frame rates etc

Choice example• Given a particular compression algorithm• IF broadcasters choose to deliver

– three different frame rates (High. Medium and Low)

– three different audio quality levels– three levels of image quality (compression rates)

• Gives 27 options - ignoring other parameters

Graphical representation

Audio Codecs

VideoCodecs

Bandwidth

Audio Compression

VideoCompression

Frame Rate

Scale Linearity• Human perception of video quality is non-linear

– E.g linear scale for video compression• 0%-25%-50%-75%-100%

• Does not provide usable choices

• More likely to conform to a power law

R=KSBR=the observer’s perception of magnitude

S= measurable stimulus,

K is a constant and

B is the compression function used by the observer

Video frame rate

• Experiments performed by West et al show B=0.68 for video frame rate

Magnitude estimation results describing the relationship between frame rate and the perception of smoothness.

0.5

0.75

1

1.25

1.5

1.75

2

0 0.25 0.5 0.75 1 1.25 1.5 1.75

log(frame rate)

log(

resp

onse

)

Content type

• Users prefer different QoS settings for different content types

• Could be related to – user’s goals or – perceptually-based– or both at the same time

Scaling Quality

• The scale used to measure QoS are a difficult area– high, medium, low are subjective judgements

(but probably the easiest to agree upon)– Scales that use more levels are used differently

by different users– QoS agreements need to be made with

individuals - who all have different perspectives

Presenting the options• Quality levels should be in the scale that the

user needs (need to determine high and low bounds)

• Content-related effects need to be considered• Quality settings should seem to be equally

spaced when viewed• Quality labelling should reflect what is to be

expected

QoS in the Interface

• QoS is dependent on the various parameters of the broadcast– this is not a simple relation– many dimensions used to provide overall QoS– e.g. audio and video compression,frame rate etc– Interface could have control over each

parameter– or some particular combinations

Comparison of QoS parameters

Use of the comparator• Right screen presents preset value - Left screen

presents users changed version

• The 2 TV model – naturally affords the process of comparison– but still makes it difficult to determine audio

parameters • (no easy comparison)• video interference (associated video events interfere)

– easier to compare video without sound!

Summary

• QoS levels can be very subjective

• Scales are non-linear but– Users tend to think of them as linear

• QoS contracts need to be based on users own perception of what QoS means

• This is different for each user!

Recommended