Introduction
Who are we?
Paul Martin
   – Started out in the industry 1996 as PlayStation programmer
   – Currently a technical director and one of the principals of Slant Six Games
   – Special interest areas; data pipelines, technical management, graphics
     rendering, special effects
   – Technical lead for SOCOM: Confrontation

Ken Felton
   – Entered the game audio world in 1994 from Film/TV/Music business.
   – Currently Sound Design Manager at Sony Computer Entertainment
     America- Foster City, CA
   – Special interest areas; Remote recording, run time audio DSP
   – Audio content manager for SOCOM: Confrontation
Introduction
This talk?
•   Audio development for SOCOM: Confrontation
•   Challenges
•   Solutions
•   Collaboration between SCEA & Slant Six
•   Surprises
SOCOM: Confrontation
Features / Specs:

• 32-player simultaneous online multi-player
• Extensive online community support
• Third-person, tactical shooter genre
• Online only
• Up to 32-player simultaneous multiplayer
      - 4 vs 4, 8 vs 8, 16 vs 16
• Up to 35 on-screen characters (32 + 3 AI)
• Large rich environments
• 7 game modes
• 3-D audio
• Voice chat
•1st title to ship with PlayStation™ Headset
Audio Technology
• Audio emitters: Any audio source
  – Static (e.g. environmental audio)
  – Dynamic (e.g. character interactions)
• Virtual Emitters: Emitter proxy for
  occluded/indirect audio path network
• Virtual Emitter network: Defines pathways
  between Virtual Emitters
Virtual Emitters
Virtual Emitters




             SOCOM: Confrontation
             Crossroads Level: Virtual Emitter Network
Virtual Emitters
Audio Occlusion
• Problem:
  – Audio filtering due to occlusion can be extremely
    expensive
     • Many ray casts!
         – Large PPU cost on PS3
• Solutions:
  – Virtual Emitters
  – Careful placement of virtual emitters
  – Batch ray casts & process on SPU
     • Latency not frame-critical for audio –can wait for results
  – Optimize code!
Audio Occlusion
• Takeaway-
  – Can be extremely expensive to implement well
  – Sounds incredible if you do it right
  – Use virtual emitters!
  – Optimize your ray casts
  – Use SPU if available!
Audio for Physics Objects
• Problem:
   – Audio simulation of real-time physics objects
      •   E.g. rolling or bouncing objects
      •   Settling sounds
      •   Audio can be triggered frequently
      •   Strategic gameplay considerations
• Solutions:
   – Count collision contact/exit points
      • Can determine rules based on this for bouncing vs rolling
      • Tunable parameters per object
   – Priority-based audio
   – Batch similar emitters based on locality
Audio for Physics Objects
• Takeaway-
  – Priority-based approach (critical vs non-critical audio)
  – Priority for gameplay always wins
  – Can get expensive in a hurry – code smart
  – Exploit locality of audio sources
Audio Build Iteration Times
• Problem:
   – Very long iteration times between builds makes sound
     design/content editing difficult, and progress hard to
     evaluate. e.g. 4 wks w/o new build during summer ‘08.
• Solutions:
   – Careful tracking of delivery items using project
     management software so we don’t lose track of what has
     been delivered.
   – Bi-weekly conference calls with sound, dialog, music, and
     developer/production staff to discuss progress and any
     changes to design or schedule.
   – Recruiting additional sound designers to play builds and
     regress implementation of delivered sounds.
Audio Build Iteration Times
• Takeaway- Offsite sound support services will
  be dependant on the game developer for
  implementation of assets. The sound team’s
  job is not complete until the assets work well
  in the game and production signs off. Shorter
  iteration loops make a better sounding game.
  Sound teams should discuss build delivery
  schedules in pre-production and have a back-
  up plan for making progress even without
  updated, regular builds.
Sound RAM reductions
• Problem:
  – Sound RAM was cut by 50% of its original size. A
    significant SRAM cut after the Beta Milestone.
• Solutions:
  – Streaming of all character grunts/dies/etc. We saved
    ~2MB of Sound RAM
  – Streaming of helicopter extraction sequences.
  – Streaming of scripted sequences when possible
  – Man weeks of careful review of all SFX samples in the
    game- delete, down sample, trim, etc.
Audio Memory Budgets
Sound RAM Reductions
• Takeaway- Plan for worst possible case RAM
  scenario – Sony Sound would have leveraged
  our streaming grain feature far more, and
  designed the sound scape very differently, if
  we had any idea that our final sound RAM cap
  would end up at 50% of the original budget.
Talk to Us

Ken Felton: ken_felton@playstation.sony.com
Paul Martin: paul@slantsixgames.com
Q&A



Any questions?

Confrontation Audio GDC 2009

  • 2.
    Introduction Who are we? PaulMartin – Started out in the industry 1996 as PlayStation programmer – Currently a technical director and one of the principals of Slant Six Games – Special interest areas; data pipelines, technical management, graphics rendering, special effects – Technical lead for SOCOM: Confrontation Ken Felton – Entered the game audio world in 1994 from Film/TV/Music business. – Currently Sound Design Manager at Sony Computer Entertainment America- Foster City, CA – Special interest areas; Remote recording, run time audio DSP – Audio content manager for SOCOM: Confrontation
  • 3.
    Introduction This talk? • Audio development for SOCOM: Confrontation • Challenges • Solutions • Collaboration between SCEA & Slant Six • Surprises
  • 4.
    SOCOM: Confrontation Features /Specs: • 32-player simultaneous online multi-player • Extensive online community support • Third-person, tactical shooter genre • Online only • Up to 32-player simultaneous multiplayer - 4 vs 4, 8 vs 8, 16 vs 16 • Up to 35 on-screen characters (32 + 3 AI) • Large rich environments • 7 game modes • 3-D audio • Voice chat •1st title to ship with PlayStation™ Headset
  • 5.
    Audio Technology • Audioemitters: Any audio source – Static (e.g. environmental audio) – Dynamic (e.g. character interactions) • Virtual Emitters: Emitter proxy for occluded/indirect audio path network • Virtual Emitter network: Defines pathways between Virtual Emitters
  • 6.
  • 7.
    Virtual Emitters SOCOM: Confrontation Crossroads Level: Virtual Emitter Network
  • 8.
  • 9.
    Audio Occlusion • Problem: – Audio filtering due to occlusion can be extremely expensive • Many ray casts! – Large PPU cost on PS3 • Solutions: – Virtual Emitters – Careful placement of virtual emitters – Batch ray casts & process on SPU • Latency not frame-critical for audio –can wait for results – Optimize code!
  • 10.
    Audio Occlusion • Takeaway- – Can be extremely expensive to implement well – Sounds incredible if you do it right – Use virtual emitters! – Optimize your ray casts – Use SPU if available!
  • 11.
    Audio for PhysicsObjects • Problem: – Audio simulation of real-time physics objects • E.g. rolling or bouncing objects • Settling sounds • Audio can be triggered frequently • Strategic gameplay considerations • Solutions: – Count collision contact/exit points • Can determine rules based on this for bouncing vs rolling • Tunable parameters per object – Priority-based audio – Batch similar emitters based on locality
  • 12.
    Audio for PhysicsObjects • Takeaway- – Priority-based approach (critical vs non-critical audio) – Priority for gameplay always wins – Can get expensive in a hurry – code smart – Exploit locality of audio sources
  • 13.
    Audio Build IterationTimes • Problem: – Very long iteration times between builds makes sound design/content editing difficult, and progress hard to evaluate. e.g. 4 wks w/o new build during summer ‘08. • Solutions: – Careful tracking of delivery items using project management software so we don’t lose track of what has been delivered. – Bi-weekly conference calls with sound, dialog, music, and developer/production staff to discuss progress and any changes to design or schedule. – Recruiting additional sound designers to play builds and regress implementation of delivered sounds.
  • 14.
    Audio Build IterationTimes • Takeaway- Offsite sound support services will be dependant on the game developer for implementation of assets. The sound team’s job is not complete until the assets work well in the game and production signs off. Shorter iteration loops make a better sounding game. Sound teams should discuss build delivery schedules in pre-production and have a back- up plan for making progress even without updated, regular builds.
  • 15.
    Sound RAM reductions •Problem: – Sound RAM was cut by 50% of its original size. A significant SRAM cut after the Beta Milestone. • Solutions: – Streaming of all character grunts/dies/etc. We saved ~2MB of Sound RAM – Streaming of helicopter extraction sequences. – Streaming of scripted sequences when possible – Man weeks of careful review of all SFX samples in the game- delete, down sample, trim, etc.
  • 16.
  • 17.
    Sound RAM Reductions •Takeaway- Plan for worst possible case RAM scenario – Sony Sound would have leveraged our streaming grain feature far more, and designed the sound scape very differently, if we had any idea that our final sound RAM cap would end up at 50% of the original budget.
  • 18.
    Talk to Us KenFelton: ken_felton@playstation.sony.com Paul Martin: paul@slantsixgames.com
  • 19.