GDC: Day 3 Notes

Day 3 Notes include the sessions, “Production Culture: How Two and a Half Artists Completed the FX for Gears of War 3,” “Everything I Know About Being an Art Lead I Learned From Rocksmith,” and “The Automation Trap and How BioWare Engineers Quality.”

Production Culture: How Two and a Half Artists Completed the FX for Gears of War 3

Production efficiency through artists point of view

8300 hand placed fx systems
2200 unique particle systems
95 mins of cinematics

2 artists for a year
Regular working days until last few days

Peers have 2-6 times larger teams

Unreal engine
Asset modularity
Easy experimentation
Fast iteration – unreal cascade tool
Visual scripting – kismet, node based scripting

Game was motivating

Production culture:

Developer qualities
Sharers not hoarders
Culture fit makes sure people can work together
Self driven without management or leads
Experience… Artists at epic would be a senior level somewhere else

Production structure
Thin hierarchy, 30 artists one lead
Role playing
Best person for the job gets to the job
Joint ownership over all assets – touch what needs work

Team building
Work in proximity
Facilitate impromptu meetings
Big family feel – epic m.o.m – lunch date and beer runs
Small size keeps it personal – even execs contribute to the family feel

Joint experience
Keep teams together
Facilitate communication – “go to” people
Self-running- especially for sequels

Inv.olvement
In decision making – good idea, do it via POC – proven ideas get not the game – no critique hierarchy- every one’sopinion counts
GameJam – contribute to the direct epic is heading- every one pitches game idaes

All this promotes ownership in the product which creates people invested in the game, their work, passion and efficiency.

Efficient tools

Everything I Know About Being an Art Lead I Learned From Rocksmith

Being a lead is hard.
– sometimes we make it harder than it needs to be

Don’t be a bitter TA

8 eng
7 audio
4 gameplay
8 env art
4 prop
5 people on mini games

Technical art project lead
-art directed game and shanghai

Process stop worrying and learn to love getting things done in a traceable repeatable way

What went wrong
Bad at juggling roles
-Failing to acknowledge doing things take time
Outsourcing became a free time thing
– never got full attention
Horrifically bad estimates
No tracking
– keep head down and keep crunching without tracking was not a good thing to do
Mini games massive time sink… Cultural gap
Keep too much work for yourself
– control issues lack of trust
Refused to ask for help until really too late
-insecurity
Everything took three times longer an estimate

Process
Do scrum agile or anything that works for you and what you believe in
Be honesty and reasonable with predictable repeatable estimates
Everyone is on the same page

Need nurturing environment
People should feel enabled
Lack of trust, scoot over I’ll do it

Hire the right people… Would you live with this person…watch out for padded resumes

When you’re stressed keep your cool

What did we learn
Act like an adult
Use process
Keep your head and don’t throw things

The Automation Trap and How BioWare Engineers Quality

Software testing through automation

Scenario
Built a bot to test all the worlds but almost daily creating a ROI black hole.

Scenario

Common assumptions
-quality comes from testing

Most devs unaware of distinction between qa and qc

The costs of testing
Testing costs= prep costs + execution costs x no. Of runs

Manual test low prep high execution
Automated= hi prep low execution

100000 hrs of automated testing does not = savings of a manual tester because the automated testwon’t necessarily be useful

Manual tester will knock-on defects
– evaluate software wholistic ally
– tests you don’t bother to do manually should be automated

Manual test
Agile adaptive
Variated
Holistic
Exploratory
Anticipatory

Automated costs
Lots of dev effort gets put to support Uto ated tests
Developing an automated framework costs developer time

Building code to test code also requires code to test the code that tests the code
Requires at least one senior engineer to oversee automation
Games change so much that it breaks the scripts beforethe tests an be deployed

New assumptions
Quality comesfrom iteration
Testing is a small cost of dev costs
Manual testing and automation re distinct
Use automation springly
Defect prevention is paramount

What makes games good?
You find more bugs than you can fix so focus n fun
Support
Initiatives to get new features in the game quicker
Som bugs are too cool to fix… Not a bug a feature

Defect prevention

Dev model
Development: Dev self test, peer review, unit test
Peline: presubmission smoke,content smoke, CIS
test: BVT, regression

Tool assistd testing can be better than automated testing.

Automation is a big investment
Set modest goals to return meaningful results
Don’t spend 8 months on automation software

Automate processes not people
Automate tasks that people cannot perform
Favor early stage dev

Drawdown age pic
B4bug pic

Test automation

Telemetry or heuristics much more useful than automated testing

Tagged

Leave a Reply

%d bloggers like this: