R

1
RD in the games industry

• Dr I Badcoe,Infogrames Development Studios
  Sheffield House

2
A talk in four parts

• Background
• Production/Development
• Mainstream research
• Further research

3
Part 1 Background

• The industry
• Infogrames
• Sheffield House
• Me

4
The Industry

• Came from Bed-sits
• Went to Many Small Independent Development
  Houses
• Now at Few Large international
  Publisher/Developers and (semi)Independent
  Studios
• Was doing Ad-hoc simplistic development by
  recognized gurus
• Now doing Increasingly sophisticated
  management/design required

5
Infogrames

• Came from Lyon
• Went to Europe
• Now at Worldwide
• Was doing PC/Handheld gamers games
• Now doing Console/Handheld
• mass-market
• licensed properties

6
Sheffield House

• Came from Gremlin Interactive
• Went to UK success, limited elsewhere
• Now at Infogrames largest and most- advanced
  internal development facility
• Was doing Consoles/PC all gamer genres
• Now doing Consoles/PC mass-market licensed
  titles

7
Me Dr Ian Badcoe

• Came from Biochemistry
• Went to Gremlin Interactive
• Now at Infogrames Sheffield House
• Was doing Head of RD Team
• Now doing Lead Programmer on Superman

8
Part 2 Production/Development

• Some numbers
• Process stages
• Environment
• Team environment
• Team structure

9
Production Some (loose) numbers

• Time Scale 18m 2yr
• Cost 1,000,000 - ?
• Production Staff 10 - 20
• Other Staff 5 10 ( shared facilities)
• Platforms 1 4 (a)
• Languages 4 7 (b)
• SKUs a b

10
Production Process Stages

• Concept genre, license, demography, platform
• Prototype gameplay, technology, feasibility,
• design
• Build make the software and artwork
• Alpha serious gameplay evaluation,
• some content still developing
• Alpha 2 functionality testing
• Beta does it really all work
• Final hardware vendor approval
• send for production
• Archive try not to lose it

11
Production Environment

• Languages C/C/ASM
• Environment PC-based.
• Console DevKit HW
• Project mgt light but formal
• Software process evolving formality
• Software design mutating,
• more reuse formality
• Testing Informal unit testing
• Formal system testing

12
Production Team Environment
Studio
Publishing Label
Studio Manager
Head of Label
Dev. Producer
Core-Tech
Publishing Producer
Marketing
Brand Manager
Other Designers
Production Team
QA
Mocap Studio
Qualitative
Sound Studio
13
Production Team Structure
Studio Manager
Production Team
Producer
Lead Programmer
Lead Artist
Lead Designer
Junior Designer
Artists
Programmers
Level Designers
14
Part 3 Mainstream research and the State of
the Art

• Hardware Support
• Animation
• Rasterizing
• Scene Rendering
• Simulation
• Sound
• AI

15
Research Hardware Support

• Hardware is continually changing
• Considerable research required to remain on the
  spot
• PCs are worse, consoles only change once every 5
  years

16
Research Animation (1)

• Body
• Hierarchic, or vertex-based
• Current games tending towards multiple bone
  weighting systems, or general morph-targets
• Mocap, or hand-edited
• Some run-time IK, not a brilliant reputation yet
• Real-time animation blending now practical

17
Research Animation (2)

• Facial animation increasingly practical
• Lip-synch
• manual
• automatic preprocess
• automatic at run-time
• Expressions
• Automatic application to different faces
• Merging the two
• Inhuman faces

18
Research Rasterizing features

• A wide variety of previously off-line
  techniques coming to the run-time
• Patch rendering
• Multitexture mapping techniques
• Shadow/Lighting volumes
• Alpha buffers
• Stencil buffers
• Hardware TL more lights becoming practical
• Procedural textures (2D)
• Anti-aliasing (various)

19
Research Rasterizing techniques

• Environment, gloss, depth, bump, , mappings
• Detail textures
• Non-photorealistic rendering (e.g. cartoon)
• Shadows, localized lights, other volumetric
  effects
• Motion blur, depth of field

20
Research Scene Rendering

• Worlds are getting more complex
• Scene-graphs are used, no standard approach
• Portal or tree-based culling
• Hang other requirements on SG, such as AI data
• Procedural geometry
• LoD techniques
• Fixed geometry LoDs
• Render-time subdivision
• View and silhouette-based

21
Research Simulation (1)

• Based on numerical techniques generally only
  Euler integration
• Generally used for rigid-body physical simulation
• Difficulties arise from
• Step-size larger than other simulations
• Presence of many discontinuities
• Approximations have non-linear knock-on effects
• Clash of physical and logical paradigms

22
Research Simulation (2)

• Requires efficient and accurate collision
  detection
• Collisions occurring in mid-simulation step
• Rotations
• Complex geometry
• Numerical precision limitations
• Software architecture problems
• Efficient call-backs to main program
• Speculative placement queries
• Free write access to simulation internals, w/o
  compromising efficiency

23
Research Sound

• State of the art
• 3D spatialised sound playback
• 32 channels
• Limited playback effects, such as reverb
• Environmental sound processing
• Interactive Music (DirectMusic, et al.)

24
Research AI

• It is very difficult to find clear examples of
  high tech AI used in games. Where it does
  occur, it is in specialized contexts such as the
  limited nn/ga in Creatures.
• However, use of low tech AI techniques is
  pretty-much endemic
• Path-finding
• Simple expert systems
• Custom rule-based systems
• Tactical decisions

25
Part 4 Advanced Research

• Animation
• Curved surfaces
• Physics/Simulation
• Sound rendering
• AI
• Network strategies
• Generalities and convergences

26
Advanced Animation

• More bones, verts, polys
• More biological control mechanisms
• Elasticity and thickness of skin
• Inherently non-squashing joints(say via layers
  of flesh)
• Attach clothing to figure animation
• Merge IK and animations to give run-time
  customization/targeting
• Better facial techniques

27
Advanced Curved surfaces

• Problems of patch topology
• Triangular patches
• Better tools
• Subdivision surfaces
• Beyond smoothness
• Overlap with animation
• Procedurally created surfaces
• (shattering, liquids, )

28
Advanced Physics/Simulation

• Better or novel integrators
• Yet to see efficient, error-free, arbitrary
  geometry, collision-detection
• Yet to see an internally consistent, usable,
  efficient physics-engine (but some reasonable
  candidates)
• Material-based physics
• Integration with animation and/or IK?(Joint
  limits, muscles, proprioceptors)
• Scale up number of objects.
• Soft objects?
• Liquids?

29
Advanced Sound rendering

• Given an environment with geometry and materials,
  should be able to render its sound properties.
• Initially preprocess approximate, big data
• Eventually could calculate on the fly(but would
  need full convolution in sound HW)
• Good for immersion
• Good for intuitive consistency

30
Advanced AI

• Higher-tech strategic AI techniques
• Bring commercial/academic AI to games, but,
• What use is that AI really?Strategies for
  visible AI assistants, god-gamesAI for
  background is arguably a waste of time.
• Behavioural AI
• More peoply people grudges, favours, obsessions
• Again, needs to be visible to be useful
• Constructive AI
• Creates the world in which play occurs
• Tunes features to player skill and preference,
  or,
• Has a personality and style of its own
• Yet again, may need to ensured its highly visible

31
Advanced Network strategies

• Distributed simulation techniques
• Resolution of desynchronized sessions
• Partially synchronized sessions
• Local server architectures
• Digicash techniques alternative uses
• True copy protection

32
Generalities and convergences

• Overall trend is from hyper-tuned, limited,
  custom code, towards efficient, more-general,
  library code, however
• Trend is very slow
• More-general is still quite specific
• Systems which set out to be too broad are
  generally rejected
• Systems which are PC only are not much use
• Anything reusable must be highly customizable
• Very little library code is from external sources
• Integration of different libraries is awkward

33
Generalities and convergences

• Techniques become available as game-platform
  power increases, but it takes time to work out
  how to use them.
• The support tools for a new technique are at
  least as complex as the technique itself
• Evolution of genres which use a technique, can
  take several generations
• Player understanding can need educating before a
  genre becomes popular (e.g. a new style of play
  or controls)

34
Generalities and convergences

• Increased power of representation (only one
  expression of which is realism), leads to
  increased immersion, and intuitive consistency
  (discuss)

35
Bonus Part Furthest Research

• Quo vadis?
• Just what is a game, anyway?
• What is fun?
• What is entertainment?
• How difficult should it be?
• When is it appropriate to play with software?
• How good are these controls?
• What is our rôle-model? Movies? TV? Books?
  Plays? Board games?
• Who are our audience(s)?

36
Further disinformation?

• ibadcoe_at_uk.infogrames.com
• (24hours hotline)