Download the full presentation here: http://www.guerrilla-games.com/read/horizon-zero-dawn-a-game-design-postmortem
Abstract: Going through early prototypes and delving into design decisions and processes that shaped development of the game, this talk gives insight into the journey game design went through while moving from an ambitious paper concept to a finished open world action RPG, with all of the small and large design decisions and choices that have to be made along the way.
4. GAME DESIGN = NOT EASY
• Concepts = Very little to go on
• Prototypes = Little polish
• Production = Not everything there yet
…Doubts… …second guessing… …negative
feedback…
26. Design Team Structure
Core Design
Player
Systems
Combat
User
Interface
World Design
Encounters Activities
Living
World
Quest Design
Main
Quests
Side Quests Tutorials
Concept Pre-Production Production Polish
31. Unique spin on World
ElementsConcept Pre-Production Production Polish
Context 4
32. Takeaways
• CONTEXT helps everything you do
• But don’t do it too early?
Concept Pre-Production Production Polish
33. Steadily finding answers…
…Combat against robots? Destructible parts, weak spots, Large encounter spaces
…Open world? Mapped out size, visuals coming together
…Player abilities? Primitive weaponry, not guns
…Story? World context established, Main character born
…World Interactions? Talk to NPCs, Merchants
Concept Pre-Production Production Polish
49. End of PRODUCTION
…Combat against robots? DONE
…Open world? DONE
…Player abilities? DONE
…Story? DONE
…World Interactions? DONE
Concept Pre-Production Production Polish
Humanoid combat?
Player pacing vs. narrative pacing?
Accessible game vs. complex combat?
I’ve been asked to remind you to please turn off CELL PHONES and don’t forget to LEAVE FEEDBACK through the FORM you get after the session.
…
Thanks so much for taking the time to come to my talk… I feel truly HONORED to be here, representing Guerrilla and HORIZON
Let’s get started!
TWO years ago, less than ONE YEAR before we were going to ship HRZ zero dawn…
We were FINALLY doing our first BIG PLAYTEST!
20 Players, 5 days, playing through the ENTIRE game, back to front, for the FIRST TIME.
We had setup telemetry and analytics, surveys for the testers to fill out and we brought our latest and greatest build…
…it was going to be AMAZING.
…But the feedback we got back, almost instantly, was BRUTAL.
The game was UNBALANCED, TOO DIFFICULT, bugs EVERYWHERE, core mechanics were NOT fun…
There’s really nothing like playtesters writing, in ALL CAPS, that your game is “beyond disappointing”……..
After FIVE years (!) of development and less than 12 months to go, I felt like we were in trouble.
I truly felt PANIC.
I think, we as GAME DESIGNERS have quite a TOUGH JOB
At the start of a project, YOU have to drive gameplay concepts and ideas with VERY little to go on
YOU have to CONVINCE your BOSSES and TEAM MEMBERS that your concepts are any good
YOU have to work hard to get parts of your designs in-game through PROTOTYPES
And as soon as you build a PROTOTYPE, the FEEDBACK starts…
Some positive, but usually mostly negative because its only 10% of what its going to be
And this all creates DOUBTS… Did you make the right choice? Did you set the right goals? Second guessing yourself constantly.
And this proceeds for the ENTIRE development of the game, it doesn’t even stop when you SHIP… If you dare to read FORUMS
TODAY… Big success… Critical acclaim, great sales… But most importantly, MILLIONS of players who loved the game we built.
So the hard work IS worth it! Sometimes YOU just forget that when you’re knee deep in development.
Now I know, it’s a Game Designers JOB to stay upright, take in the feedback, analyze it and try to TACKLE all those PROBLEMS head on…
And that’s what I want to talk about today… I want to share OUR VERSION of staying upright.
I want to take you through the journey that the GAME DESIGN DEPARTMENTS at GG went through, making HORIZON
I want to take you through EACH PHASE of development and talk about…
The INTENTIONS we had, the GOALS we set, what we tried to ACHIEVE
The PROBLEMS we faced… Not ALL… But the most interesting ones…
The REFLECTIONS we had THEN and what we tried to FIX… But also AFTER SHIP, looking back now
My hope is that this can INSPIRE you to PRESS ON when things get bad and that maybe you can LEARN from our mistakes.
Hi, my name is Eric, and I am one of the LEAD DESIGNERS at Guerrilla Games.
I’ve been at the company for almost 15 years and have worked on almost every game Guerrilla has shipped.
I started out as a single player level designers on KZ1, moved into game design on KZ2, went from SP to MP/Online and eventually became LEAD on KZ3 and KZSF.
I’ve been trying to stay UPRIGHT for quite a few projects and have seen my share of PROBLEMS… And it only gets little easier over time
JOURNEY STARTS 2011
EARLY CONCEPT PITCH
FOUR MAJOR GAME DESIGN GOALS
EXPERIENCE the MAJESTIC WILDERNESS of a FUTURE POST-POST APOCALYPTIC WORLD
Make a BEAUTIFUL world where the players WANTS to be
<killzone opposite example>
RULED by AWE-INSPIRING MACHINES
UNIQUE combat, something NEVER done before
INHABITED BY INTERESTING AND EXOTIC TRIBES
Humans go EXTINCT
What would come back? How would those cultures affect a Player Experience?
Open World = SYSTEMIC GAMEPLAY = Potentially ENDLESS
But… KILLZONE experience only
Knew that it is an ENORMOUS CHALLENGE
INTRO:
Started with CONCEPT phase
Ended up taking 2.5 YEARS – Long time but…
SMALL and MULTI-DISCIPLINARY team, REST OF TEAM on SHADOW FALL
GOALS:
DESIGN what Horizon will be, but more importantly
MAKE IT PLAYABLE
Find out WHAT WORKS and WHAT DOESN’T
PROBLEM:
CONCEPT TEAM = 8 TO 16
Designers, Artists, etc.
Using KILLZONE content/engine/tools
Autonomous to build prototypes
PLAN/APPROACH:
Answering HIGH LEVEL QUESTIONS
Machines = KEY USP… But how EXACTLY?
What makes a HORIZON open world? What makes a GUERRILLA open world?
What kind of ABILITIES, what is the STORY we want to tell, how do you INTERACT with the world?
PROCESS:
GOING WIDE = CORE
PLAY ideas, not just DESIGN
Find what WORKS, remove the REST
Now will go through some EXAMPLES of PROTOTYPES
EXAMPLE QUESTION 1 – COMBAT
INTENTION: ANSWER combat QUESTIONS (how does combat against robots work exactly?)
PROTOTYPE: Try to build the BIGGEST MOST COMPLEX ROBOT first
REFLECTION: EARLY DESIGNS SPOT ON – Vulnerabilities, weakspots, armor, use weaponry against it, etc.
EXAMPLE question 2 – OPEN WORLD
INTENTION: ANSWER open world QUESTIONS (What can you do in our world? What kind of interactions? Etc.)
PROTOTYPE: Try to build the BIGGEST MOST COMPLEX SETTLEMENT first
REFLECTION: PLAYGROUND to test density, size, feeling, looks, atmosphere, interactions
EXAMPLE question 3 – PLAYER ABILITIES
INTENTION: ANSWER ability QUESTIONS (What kind of weaponry? How do you traverse this immense world?)
(one of the) PROTOTYPE(s): Focus on challenge for MOUNTED traversal
REFLECTION: Never did it before, had to really LEARN how MOUNTED TRAVERSAL worked from a game design perspective
TAKEAWAYS:
This type of PROTOTYPING really helped make HORIZON what it is today
But its NOT CHEAP
Lots of prototypes REDONE, THROWN AWAY, etc.
It essentially TAKES a lot of TIME / EFFORT
PROBLEM LATER ON:
Not a COHESIVE EXPERIENCE
Scary for team! Doubts, second guessing, etc.
Great ISLANDS… but where is the GAME?
INTENTION/GOAL:
Create the COMPLETE GAMEPLAY LOOP in a POC
Try to SEE EVERYTHING WORK TOGETHER
Explain LOOP – EXPLORE (walk, traverse, climb), INTERACT (talk to NPCS, hunt ROBOTS), GATHER (robot parts, quest rewards), PROGRESS (upgrade weaponry, armor, skillset)
REFLECTION:
CONFIDENCE BOOSTED!
EARLIER NEXT TIME?
COULD WE have done that?
TAKEWAYS:
GOING WIDE / ISLANDS = allows you to find WHAT your game is
BUILDING FULL LOOP = allows you to DETRACT what your game is NOT
END OF CONCEPTING:
PARTS of questions being answered
Some questions NOT ANSWERED AT ALL
NEW questions being raised…
And in 2014, at the END of the concept phase, we SHIPPED KZSF… The team effectively HAD to move over…
INTRO:
Even though CORE PLAYER SYSTEMS were not fully defined, having more designers allowed us to ALSO start focusing on WORLD SYSTEMS like QUESTS, ACTIVITIES, things you would DO in the world
Having more members of OTHER DISCIPLINES allowed us to FOCUS on BUILDING THE WORLD (art assets, characters, terrain, foliage, etc.)
Plus it allowed us to start focusing on STORY and NARRATIVE as well
INTENTION:
MORE DESIGNERS with SPECIFIC FOCUS
Explain team structure/responsibilities
PROBLEM (but good in the end):
We had a lot of story IDEAS / rough thoughts (example?)
Writing was done by the concepting team, no dedicated team members
Lacked CONSISTENT CONTEXT for everything we were doing
SOLUTION:
DEDICATED NARRATIVE team formed
First thing they did was LOOK AT WHAT WE HAD and changes MAJOR PARTS
Felt bad at first, but what they did very well is establish CONTEXT for EVERYTHING WE WERE DOING
CONTEXT Example 1: ROBOT RIDING
PROBLEM:
We wanted the player character to be special
We wanted the player to interact with robots
Big animals were getting in the way
SOLUTION
Remove big animals
Make the player able to ride ROBOTS
Achieved both goals, and less work
CONTEXT Example 2: MAIN CHARACTER
PROBLEM:
Aloy was always there or at least BITS of her
But it lacked details
Without those details, you cannot truly DEFINE what she CAN DO
SOLUTION
Define CHARACTER, ARC, PERSONALITY, STYLE
Allowed us to FOCUS the gameplay AROUND her
Made everything feel like it had a PLACE
CONTEXT Example 3: ROBOTS THEMSELVES
PROBLEM:
We knew what INDIVIDUAL robots could do from a COMBAT perspective
But we didn’t yet understand why robots were there CONTEXTUALLY
Without that, they lacked the right PERSONALITY or FEEL, when not engaged in combat
SOLUTION
Define reason why ROBOTS are there (terraforming, rebuilding the planet, part of a bigger plan etc.)
With that CONTEXT, we could better define the LOOKS, the BEHAVIOR and the FEEL of robots
It also generated a LOT of ideas for MORE robots
CONTEXT Example 4: HORIZON WORLD UNIQUENESS
PROBLEM:
We knew what kind of activities we wanted
Similar to existing open worlds (underground dungeons, challenges that unlock parts of the map)
But they did not feel HORIZON at all
SOLUTION
Defining the context for the world, WHY THE WORLD IS THE WAY IT IS, allowed UNIQUE SPINS on existing activities
“Challenges that unlock parts of the map” became TALLNECKS, underground dungeons became CAULDRONS
Even though the gameplay was the same, CONTEXT made our world feel UNIQUE and WONDERFUL
TAKEAWAYS:
CONTEXT makes everything better
It allows you to focus which MECHANICS work and which don’t
It allows you to define your WORLD better and make it UNIQUE
REFLECTION:
But being able to TRY OUT THINGS without CONTEXT first, is equally important!
If you add CONTEXT too soon, you might prevent very cool ideas to be born…
END OF PREPROD
Worked out how STORY and WORLD worked more
The GROUNDWORK for HORIZON was now there: CORE MECHANICS + WORLD + NARRATIVE
But this is where the bigger PROBLEMS started rearing their ugly head…
INTRO:
We were now able to start on building our ENTIRE game
We felt we had a lot of the CORE MECHANICS (not all) but that we had ENOUGH to get started
PROBLEM:
However, our INEXPERIENCE WITH OPEN WORLDS started showing here
Want to talk you through SEVERAL PROBLEMS we ran into at this stage
NOT ALL, but I cherry-picked ones you could LEARN from
INTENTION OVERALL:
We wanted to START AT THE BEGINNING of the game
100% of players SEE this part, HAS to be AMAZING
If you do this right, you ENTICE PLAYERS to keep on playing
INTENTION – FIRST QUESTS OF THE GAME:
GD wants to expose/tutorials CORE MECHANICS (health systems, combat, stealth)
Write them into the NARRATIVE, expose them EFFECTIVELY
PROBLEM:
QUALITY narrative costs a lot of effort (dialogue, characters, cutscenes)
Killzone = LINEAR set pieces
Horizon = Know EXACTLY which MECHANICS you need
We had to guess!
REFLECTION:
Gameplay weight SHIFTED throughout dev
Systems we THOUGHT were going to be important became LESS IMPORTANT and vice versa
Adding NEW MECHANICS later on, SHIFTED the GAMEPLAY WEIGHT of others
EXAMPLE
We thought that DISTRACTING ROBOTS (distraction rocks) and knowing their PATROL PATHS was going to be important
When we BUILD THE WORLD and the encounter spaces were MUCH BIGGER, this became less important as the player could just ‘avoid robots’ all together
TAKEAWAYS
ITERATING on NARRATIVE is NOT EASY if you want to achieve HIGH QUALITY
So was starting at the beginning a good thing?
I would argue that starting on the beginning WHEN YOU KNOW YOUR ENTIRE GAME is better
INTENTION:
Horizon = MULTIPLE SOLUTIONS possible = YOU CHOOSE
There is no PERFECT solution
BUT Horizon = ACCESSIBLE as well
PROBLEM:
Playtesting showed DIFFICULTY SPIKES
Players who were LESSER SKILL or LESSER ENGAGED ran into issues
Basic Arrows CAN kill something, why would they TRY SOMETHING ELSE?
REFLECTION:
How do we ENTICE players to engage with other options?
How do we do this WITHOUT losing our core pillar of there not being a PERFECT SOLUTION
REFLECTION:
We considered making specific SOLUTIONS MANDATORY
FEEDBACK = NOT POSITIVE = FEEL FORCED = NOT HORIZON!
So instead, we decided to SHOW as many OPTIONS as POSSIBLE
SOLUTION:
ALL IN on FOCUS
OPTIONAL tool = Good = Before AND during combat
NEW PROBLEMS:
Don’t want to block PRETTY WORLD
Not a lot of HUD space?
INSPIRE options, not GIVE them
REFLECTIONS:
FOCUS shows POSSIBILITIES, not SOLUTIONS
But it didn’t SOLVE the entire issue, it simply ADDED AN OPTIONAL LAYER to people that WANTED IT
It HELPED with the difficulty spikes but it didn’t FIX the core problem
TAKEAWAYS:
If you wanting the Player to CHOOSE their own solution, you have to consider HOW you’re going to EXPOSE them to the options
But you have to try and do this in your CORE GAMEPLAY, rather than us, in an ADDITIONAL OPTIONAL LAYER
Enable the player to SEE the options, without having to do ANTYHING extra
INTENTIONS:
ROBOTS = CORE
We wanted to have HUMAN combat, but as a SIDE ACTIVITY
But we started introducing more and more humanoid combat… WHY?
PROBLEM 1:
STORY without HUMANOIDS = impossible… Cannot SPEAK
So we introduced more and more human characters
Which made it go from a secondary activity to PRIMARY activity
PROBLEM 2:
We really did achieve making ROBOT combat deep and interesting
But this HIGHLIGHTED the fact that human combat was NOT
Fewer weakspots, tactics, choices, options…
SOLUTION
We wanted the HUMAN COMBAT tactical depth to come from ENCOUNTER design
Adding COVER, different APPROACHES, gaining an ENVIRONMENTAL ADVANTAGE
And to an extent, you can see this working in the CAULDRONS/BUNKERS and the BANDIT CAMPS
REFLECTIONS
Horizon = OPEN world
You therefore cannot FULLY rely on ENCOUNTER design (different approaches, etc.)
We underestimated the impact of this
TAKEAWAYS
In hindsight, we should have made SINGLE HUMAN should be AS DEEP as ROBOTS
Focus on WEAKSPOTS, BEHAVIOR, WEAPONS and ARMOR
THIS = SYSTEMIC GAMEPLAY
Less relying on encounter design
END OF PRODUCTION:
Found a lot of answers to the questions we asked
But we found out the hard way that we didn’t ask ALL the questions
Questions in hindsight, we SHOULD have asked way earlier
INTRO:
We now could play the ENTIRE game from start to finish
All content and mechanics were IN
We started PLAYTESTING a LOT
PROBLEM:
But now the META ISSUES started showing
Problems we could not spot before because we COULDN’T PLAY the entire game
Now that players were playing for DAYS and DOZENS of HOURS we were able to, for example, PROPERLY STRESSTEST our ECONOMY
INTENTION:
Our intention with the economy was to make it INVOLVE ALL CONTENT
The player would have to hunt ALL ROBOT TYPES and engage in ALL ACTIVITIES to unlock everything
This to us was what it meant to BE AN OPEN WORLD
PROBLEM 1:
This however did end up resulting in there being a LOT of DIFFERENT TYPES of ECONOMY resources
Each ROBOT had their own set, plus WILDLIFE, plus HUMANS, plus FAUNA
And the PLAYER was having difficulty understanding WHAT was used for WHAT
PLUS where do I get RESOURCE Y?
INTENTION 2:
TRIBES ARE PRIMITIVE
They wouldn’t understand what a CPU is
LENS versus EYE, COIL versus BATTERY
PROBLEM 2:
Added abstraction layer = Added confusion
Player can also not understand VALUE of something due to AMBIGUITY
SOLUTION:
Late to TRULY change this
Add TOOLS to MITIGATE problems
TRADER JOBS
REFLECTION:
Too deep, not core
Didn’t fix core problem
TAKEAWAYS:
We should have realized the economy would be TOO COMPLEX earlier
Even though we couldn’t TEST it, in hindsight, we SHOULD have realized there was a problem
If one of your CORE PILLARS is ACCESSIBILITY, making sure your ECONOMY is simple but effective is key
OUTRO:
GAME LAUNCHED
DIDN’T DO TOO BAD
MOST SUCCESSFUL GAME WE HAVE BUILT TO DATE
So again, all the HARD WORK is worth it…
FINAL TAKEAWAY:
But if you take anything away today, it’s this…
REVIEW often… Review your GOALS, your INTENTIONS, your PROBLEMS, constantly reflect during development to flag issues as early as possible…
Be HONEST with yourself; what WORKS what DOESN’T… Don’t get STUCK in ideas that are not working out
And get READY for the problems; build in FLEXIBILITY in your processes so you CAN ADAPT if necessary
If you FACE your problems HEAD on, it’s much easier to STAY UPRIGHT…