Docstoc

GDC 2005

Document Sample
GDC 2005 Powered By Docstoc
					Successful Outsourcing on AAA Games
- A Case Study of Forza Motorsport™

 Introductions:
   John Wendl-Art Director-Forza Motorsport™, Microsoft
     Game Studios™ (Redmond, WA)
   Rajesh Rao-CEO, Dhruva Interactive (Bangalore, India)
   Nick Dimitrov-Business Manager-Forza Motorsport™,
     Microsoft Game Studios™ (Redmond, WA)
The Game: Forza Motorsport™
   AAA franchise that MGS developed and published on Xbox
   The ultimate racing simulator—93% average review
    scores on Game Rankings
   Over 230 fully licensed vehicles ranging from “tuner” to LMP
    race cars
   Unprecedented levels of customization, personalization and
    damage in the sim racing space
   Here is what the press had to say specifically about the graphics
    in Forza Motorsport™:
       “Gorgeous is the only appropriate way to describe Forza’s
          graphics”—Yahoo Games
       “The graphical beauty found in Forza starts with the big, bright car
          models”—Game Spot
       “The lush graphics and usage of anti-aliasing make the “other”
          racer look Spartan in contrast”—GamePro
Making the Decision to Outsource
   MGS management initially drove the initiative to outsource
   Very large amount of content. We needed to build 18 complete
    track environments & over 230 cars in 12 months that included:
        • A very high level of detail
        • 5 unique static LOD’s
        • Damage & mesh textures
        • Upgrade body parts
        • Customizable paint/liveries
   We needed a cost effective way to dramatically increase team
    size during production
Deciding What to Outsource
   After careful consideration of our content needs the decision
    was made to outsource vehicle production primarily for these
    reasons:
       • Vehicles were the largest part of the art production
         schedule and our biggest potential schedule risk
       • Our licensed vehicle assets require little art direction
       • Compared to other assets such as track environments, there
         were fewer dependencies on our build and tools to deal with
       • Vehicles are modular components that don’t have to
         integrate with other assets

   Deciding what to outsource is a critical decision for every
    production team to make when considering outsourcing.
Vendor Selection & Due Diligence
   Vendor selection began nearly a year before actual content
    production. We looked at around eight companies all over the
    world. Many things were considered including:
       •   Team structure/depth
       •   Production bandwidth
       •   Experience with specific types of assets
       •   Cost
       •   Proximity
Vendor Evaluation & Stress Test
   We selected 5 companies to do an initial evaluation/stress test
    of 2 vehicles w/no time limit. We wanted to Compare apples to
    apples so everyone got the same test.
   The 1st evaluation/stress test consisted of 2 cars with
    essentially no time limit. Essentially we wanted to determine if
    the vendor could hit the quality bar required. We looked for:
     • Quality
     • Innovation
     • Ability to work accurately with a variety of reference
       material
   We selected 3 finalists to do a 2nd evaluation/stress test
    (speed/adherence to a spec) of 4 vehicles in 4 weeks.
    Essentially we were trying to determine their capacity. We
    looked for:
     • Speed
     • Adherence to the spec
     • Bandwidth
Making the Final Vendor Selections
   Through our extensive evaluation process (3-6 months) and
    solid metrics clear winners emerged.
   To better bolster our team and diversify our risk we decided to
    work with the two vendors on cars that best met our
    requirements.
   We were able to mitigate risk to a large degree because our
    selected vendors proved they could do the work we needed to
    the quality bar and within the time constraints required.
Preparing for Outsourcing
   Several key things needed to happen and/or be in place before
    we could begin production with an external vendor:
       • Dedicated internal staff—2 content coordination, 3 QA
       • Prototype build—this allowed us to finalize our spec &
         provide sample assets built to that spec
       • Detailed specs—our specs had to be done to a very high
         level of detail to clearly communicate our needs to the
         vendors
       • Small percentage of vehicle production done in house
       • Detailed contracts in place
       • Signed NDA
       • Sufficient lead time for vendor to allocate and commit
         resources
Production
   Key components of our external production plan include:
     • The vendors were responsible for the entire asset in build
     • Vendors were equipped w/complete art pipeline, tools and
       dev kits
     • QA was done at both the vendor and the developer
     • Assets not meeting necessary quality, accuracy or adherence
       to the spec were sent back with detailed feedback clearly
       illustrating the problem
     • Changes to spec/process
     • Adding/removing features/asset classes
     • When to amend the contract
     • One developer visit to outsourcing company during first third
       of production
     • Assets locked down and approved by manufacturers in phases
   What we did ourselves
     • Late changes to spec that affected all cars
     • Small tasks that weren’t worth the overhead of outsourcing
Shutdown
   The vendors were responsible to get all the assets completed
    and past our internal QA by our internal content complete date.
    We didn’t really have vendors doing much bug fixing after that
    but are looking to do it much more extensively in the future.
   It is ideal to utilize the manpower of an external vendor during
    this critical phase in production and shut the project down. This,
    of course, has to be weighed with the risk of introducing new
    bugs.
Lessons Learned
   What went right:
      •   Great looking results. It’s virtually impossible to tell
          which cars were outsourced and which were done
          internally.
      •   Give and take on both ends
      •   Healthy working relationships that are still thriving and
          growing
   What could have gone better:
      •   Stable pipeline, toolset and graphic feature set
      •   Bug fixing
      •   Late changes to spec
      •   Better estimation of parts work
Do’s & Don’ts in Successful Outsourcing
Do carefully screen each vendor.
Do carefully consider the specific work to be outsourced.
Do make sure detailed contracts w/milestone descriptions are in place before work
    begins.
Do take special measures to deal with working remotely across time zones &
    cultural/language barriers.
Do setup vendors with the same tools and pipeline you use internally whenever
    possible.
Do make vendors responsible for as much of the asset integration as possible.
Do create detailed, objective feedback that leaves little room for misinterpretation or
    miscommunication.
Do make the decision to outsource at the beginning of planning & pre-production.
Do pay attention to questions, queries and feedback from your outsourcing partner.
    They may see something you missed.
Don’t wait until you’re behind in your schedule to decide to outsource.
Don’t treat the outsourcing company like just another team member.
Don’t just fix incorrect assets internally, send them back with detailed feedback so
    the vendor can learn and improve future deliveries.
Don’t under communicate with your vendor. There is no such thing as too much
    communication.
Don’t assume the work of one artist is representative of the vendor’s entire
    production staff.
Don’t make the client/vendor relationship adversarial.

				
DOCUMENT INFO