Dissecting our Baby: Auto Assault Postmortem the good, the bad, and all the ugly'

1
(No Transcript)
2
Dissecting our Baby Auto Assault Postmortem
the good, the bad, and all the ugly.

• Scott Brown,
• President , NetDevil LLC.
• Hermann Peterscheck
• Producer, NetDevil LLC.

3
Overview

• Early game concept.
• Getting a publishing deal.
• Structure of the deal.
• Problems and potential solutions.
• Building the game.
• Prototype
• First Level
• Features
• Production
• Shipping
• Release
• Post Release
• Q A

4
Early Concept

• Diablo in Cars
• Diablo was fun and sold well.
• We were fans of Autoduel.
• Original game idea
• Avoid competing in flooded fantasy market.
• Create a game for a large underserved market
• Post-apocalyptic.
• People who dont like fantasy games.

5
Early Concept
Autoduel Origin Systems (1985)
Diablo 2 Blizzard Entertainment (2000)
6
Early Concept
Early Visual Representation of Auto Assault
7
Getting the Deal

• Existing relationship helped.
• Go to Conferences and meet other devs.
• We had met and knew many of the executives at
  NCSoft.
• Try and have something you made to show.
• We had already made, shipped and shown Jumpgate
  to people who were now at NCSoft.
• Strong visuals sell the idea.
• Graphics are the first thing people see.
• We didnt have a demo, but we have a concept shot
  that captured what the game would be.
• Pitch the right idea.
• People at the publisher wanted to make a car game.

8
Structure of the Deal

• Development paid by Publisher
• Most devs dont have millions.
• Dev cost is advance against royalties.
• Payments are made based on meeting milestones.
• When royalty payments exceed costs, developer
  begins to make money.

9
Structure of the Deal

• Example
• 40 people for 4 years
• Costs around 3-5 million/yr.
• Total cost of 12-20 million.
• At 10 royalty, game must generate 120-200
  million before developer makes money.
• Slipping 1 year adds another 30-50 million.
• 100k users at 15 18 million/yr so it takes
  about 6-10 years at that rate before costs are
  recouped.
• Not a good retirement plan!

10
Problems with that Structure

• Developer is motivated to meet deadlines above
  all else.
• Can lead to conflict with what is best for the
  game.
• Delays can significantly effect time to
  profitability.
• Delays happen at the end most expensive time.

11
Problems with that Structure

• Missing deadlines has consequences.
• Late payments puts stress on owners.
• Trickles down to developers.
• Dev studio cant cover costs anyway unpleasant
  situation.
• Publisher may feel forced to reward failure.
• Interrupting visits.
• Publisher checking in forces devs to create
  demos which may or may not be beneficial.
• Dread of cancellation.
• Yikes!!!

12
Potential Solutions

• Publisher and Developer have same goals.
• Make great game that sells.
• Have realistic schedule.
• Communication is key to success.
• Developer and publisher must both know the real
  state of the project.
• Frequent updates.
• Weekly builds.
• Video conference.
• Dont BLACK HOLE the development!

13
Potential Solutions

• Change Good!
• Dont make a detailed schedule and stick to it no
  matter what.
• Make list or required for ship features.
• Make game playable fast.
• Dont be afraid to change.
• Dont be afraid to change a lot.
• Make your payment schedule reflect this.
• Payment for specific features on deadline is not
  realistic.
• You dont know what you are going to be doing 2
  years from now.
• You probably dont know what you will be doing 2
  months from now.

14
Potential Solutions

• Publisher understood these things.
• Producer visited studio at least every few
  months.
• Gave advice and direction.
• Measured progress and gave vital feedback.
• Problems
• When issues occurred, there was a disconnect
  between finance and production.
• Renegotiate contract giant pain.

15
Prototype Phase The Good

• Concept Art to work from and towards.
• Focus on major issues
• Driving.
• Loot.
• Killing.
• Bare bones editors.
• Use 3rd party software.
• Renderware and Havok.

16
Prototype Phase The Good

• Avoiding competition.
• Not character based.
• Not fantasy environment.
• Physics is cool!
• Watching things move around in a physical world
  is cool.
• Blowing things into little bits is cool.

17
Prototype Phase The Ugly

• Problems with familiarity.
• Problems with identifying with vehicles.
• Does driving shooting skills fun?

Armor!
Umm Armor??
18
One Great Level The Good

• Focus on small playable area.
• Improvement on toolset.
• Establishing production chain.
• Identifying major features.
• Object model and server architecture established.

19
One Great Level The Bad

• Losing sight of accessibility.
• Difficult to evaluate how others will play our
  game.
• Evaluating quality in terms of improvement.
• Better is not the same as Great.
• Not fantasy environment.
• Top down vs. First Person
• People wanted to see the horizon.
• Substantial implication on assets and
  functionality.

20
One Great Level The Ugly

• Interest in Characters.
• Feedback from publisher/players was a need for
  character advancement.
• Started designing 12 characters instead of focus
  on one.
• Driving control.
• Feedback from publisher that game had to be mouse
  controlled (i.e. without keyboard).
• Started working on many cars instead of making
  one great.
• placer

21
One Great Level The Ugly

• At the end of this milestone we didnt have one
  great level.
• Pressure of feature development schedule for paid
  milestones.
• Fallout from major tech changes.
• Fear of project being cancelled.
• Project moved to next phase instead of fixing
  issues. placer

22
Major Features The Good

• Dynamic item generation.
• Tons of items and most are unique.
• Prefix, suffix, random stats.
• Quest driven world.
• Always something to do.
• Similar to World of Warcraft and EQ2.
• Fast paced game play.
• Not a lot of time walking around.
• Fast pace of kill-gtloot.

23
Major Features The Bad

• Features were implemented on time and within
  milestone list
• leaving very little time to play the game.
• Complex skill system.
• Designed too many skills to fill levels instead
  of what we barely needed.
• Not fantasy (unfamiliar) environment.
• Fireball vs. Hazardous Cleansing
• Complex trigger system for Quests.
• Gave designers great freedom.
• Caused massive debugging tasks.
• Lack of tool support for designers.

24
Major Features The Ugly

• Physics simulated on the server.
• Pushed havok to its limits.
• Lots of time spent making physics fast enough.
• Destructible environment.
• Artists had to create base asset and all the
  pieces.
• Assumed that once features are done we would
  polish at the end.
• There is no polish at the end.
• Polish as you go instead.

25
Major Features - Lessons

• Big red flags.
• Developers dont play game because they are too
  close.
• Eat your own dogfood!
• People say the game is good, but not my kind of
  game.
• Really? Whose game is it then? If you cant find
  audience, who will buy your game?
• Bad frame rate bad game.
• Rules are not universal.
• Different systems can be used at different stages
  of the game.

26
Major Features - Lessons

• Support during the day, development at night
  problems!
• Programmers had to support Art and Design during
  the day.
• Implemented features (i.e. hard programming) when
  tired.
• Guess what that does to feature quality?

27
Content Development

• MMOs are BIG!
• Content required lots of designers, artists,
  writers, QA, etc.
• Game should be done, including major features,
  tools, content pipeline and deployment. They
  werent.
• As we added content it had to be done and redone
  as problems were addressed.
• Resulted in missed milestones.
• This added even more pressure.
• Major extended crunch times.
• Fear of cancellation.

28
Content Development

• Game was delayed
• Publisher did not cancel game.
• Refocus on game.
• Small teams which focused on specific areas and
  dealt with specific problems.
• Began to cut content/features (whew!).
• Momentum leading up to release decreased.
• Press lost interest.
• Beta community deteriorated.

29
Content Development

• Results were positive.
• Game got better.
• Major bugs were dealt with.
• Server and client stability much improved.
• Other problems persisted.
• Still issues with performance.
• This can not be ignored!
• Balance issues.
• Some feature creep.

30
Beta Test

• What is beta for?
• Beta is not a test for functionality.
• Beta is marketing, if you like it or not.
• Beta is for stress testing with lots of people.
• Beta too early bad press and players trashing
  your game.

31
Beta Test

• Auto Assault was in beta too early.
• Game was judged as complete.
• Older screenshots are shown to this day.

32
Release

• Great sighs of relief.
• Launch was relatively smooth.
• Usual day one patch.
• Started with too many shards.
• Had 8, should have started with less and scaled
  up.
• Game seemed empty.
• Game did not meet expectations.
• Created internal stress as to future of company.
• Created stress between developer and publisher.
• Relationship stayed positive as we collaborated
  to deal with issues as they came up.

33
Post-Release

• Dealing with balance issues.
• Constantly changing skills and items.
• Deal with removing exploits that small population
  uses vs. changing something most people enjoy.
• Having a good partner.
• NCSofts experience and infrastructure ensured
  good customer service, server monitoring and
  support.

34
Post-Release

• Stat-tracking.
• Important to know where people are playing and
  where they quit.
• Used this to focus on areas where people quit the
  game.
• What people say they want and what they play are
  different.
• Lots of small patches vs. large patches.
• Important to fix problems for new players.
• Have to give existing players something new.
• (empty)

35
Conclusion

• Auto Assault was a large and complex game. We
  learned a lot!
• Maintaining a good publisher/developer
  relationship is important.
• Treat innovation with caution.
• Play your game early and often.
• Polish as you go.
• Avoid complexity.
• Expect change and plan for it.
• Focus on one good instance first, then add
  content.

36
Questions?