www.chadbordwell.com NOTE: My opinions do not represent those of my employer.

Friday, May 12, 2017

My Quest For AI Individuality

I am not a programmer, I am especially not an AI programmer.  I am a Designer so my job is to have visions and plant them in others….well sort of.  But a game did this to me and I’ve had a hard time shaking the curiosity of what if even after all these years.  I don’t even know if it was a real moment but it felt like one and that’s what’s important.

The game was Unreal.  The first Unreal.  The single player game.  I was running from a Skaarj which was a predator-type creature and common enemy in the game.  I ran into a room and dove under a stairway to at least give me time to collect myself.  The Skaarj ran into the room and stopped.  Crouched under the stairway I was watching him assuming he would spin around and shoot me using his “knows where the player is at all times” AI knowledge, but that’s not what happened.  Instead he stood there, looking around.  Apparently coming to the conclusion that he had lost me he slowly walked out of the room.  This may not seem like a big thing but this was 1998.  Back then AI ran at you and then shot at you, and not much else.  I never saw an example of this happening again in the game and even replaying it I never had it repeat.  What the heck happened at that moment?  Was it a bug? A happy glitch?  Whatever it was it left an imprint on me and my approach to AI scripting. 

When I was on the Quake 3 team I spend several weeks at id Software working as a preliminary production tester at both the beginning and end of the project often focused on game play rather than bugs.  One of the AI programmers there (Mr. Elusive) and I got into a discussion regarding Sarge’s level.  You were in a one on one match with Sarge and when he was being balanced I said it felt like he was cheating.  It felt like he was taking aim on me through the walls and shooting me the very moment I came around the corner.  Mr. Elusive assured me that it wasn’t happening, that’s not the way the AI works.  I said but it feels that way and he again said that’s not what happens.  So I explained, “Whether or not it is happening doesn’t matter.  What I’m saying is it feels like it is.  The players will feel like it’s cheating.”  That is what AI is all about, feeling correct.

In 2001 I worked on Return to Castle Wolfenstein and strove to make each AI feel unique, like they were an individual.  Through the script I was able to make some become more aggressive when their health dropped.  I made some cower and run for cover when their captain died.  I wanted them to feel real like had a sense of self-preservation mixed with a personality.  I’m not sure it was achieved but again things were pretty basic back then.

My next opportunity to put a stamp on AI was on Quake 4 but thing had changed.  Game play structures had become more systemic.  While Quake 4 AI was certainly not systemic there was no room for tinkering in the AI structure.  AI was setup to be predictable.  The guys with green armor always throw grenades, the guys with red armor have more health, etc, etc.  The archetypes were different but the individuals were not.  Once Quake 4 shipped I became a little obsessed with a Quake 4 mod I wanted to create called Grunt Hunt.  In Grunt Hunt it was you against 1 Grunt, but instead of a mindless charging creature he would run from you, pick up health and ammo drops, he would strive to live rather than be mindless fodder essentially following the same rules as the player.  Seek better weapons, ammo for weapons, and health.  However unlike multiplayer bots the pace would be slower, with faster death and higher stakes for sudden moves.  I wanted to AI to experience the world as the player did.  The player doesn’t know a heath pack is in a room until he walks in it, and I wanted the AI to not know either.  A health pack enters their field of vision, they assess their health and determine if they need it or not.  Their health drops dangerously low, they change tactics to a more hyper defensive mode, going cover to cover in search of health over attacking.  At least this was my vision.  I was quickly told by programmers this couldn’t be done.  They said the processing power needed to have an AI running around assessing their environment was too great.  So that sucked.  

On later titles I conceded to simply making the AI appear smart however since coming to Ubisoft they use a fully systemic system that designers have little control over.  I can't really talk too much about the systems Ubisoft uses but I think there is room for improvement that can rise the average AI about the rank of fodder.


AI has become more of a person pursuit at this point.  I have many titles rattling around in my head with unique AI problem to solve.  I would still love a chance to tinker around with AI ideas that go beyond the norm.

Monday, March 27, 2017

DOOM Documentary

Just came across this really good documentary about the creation of the latest DOOM.  It show the development struggles id Software had with making a new version of a legendary game.  Certainly similar struggles we had while making Wolfenstein and Quake 4 when I worked at Raven Software. Story was not id's strong suit and I'm glad they finally found a way around it that fits them.


Friday, October 14, 2016

Return To Castle Wolfenstein Is On Sale

Yes the first game I worked on as a designer is now on sale on Steam.  Maybe it's always on sale I don't know but this was the first time I noticed it at least.  If you're into retro gaming it's a really great old school shooter.  It's still one of the highlights of my career.

I learned a lot working with Gray Matter on it.  I was the last designer Activision had on the payroll when they stopped internal development to focus on publishing.  Gray Matter needed help so I was sent over to help them stay on schedule.  The schedule was an absolute death march, I started overtime in January and didn't see the light of day until November.  I literally missed the entire summer.  I slept on the floor of my office so often that Gray Matter bought me a pillow and sheets.  No joke.

The development process was much different then.  You didn't have tasks really, you were given levels and it was your job to see them to completion.  Your task was to "make it fun." Every day the company owner and design lead would play the level and give you feedback which you iterated on.  Often you were totally free to try anything you could think of if you thought it would make the game fun.  One day I hand scripted every movement of a head bouncing down the stairs (there was no physics) because I thought it would be a nice touch to a spooky area.  In modern game development you'd never really be allowed to take time to do such a thing.  You are given tasks to complete like an assembly line.  Game development today has lost most of that creative freedom.

So it was a terrible death march of endless work but it was also creatively free and fun.  In the end Activision bought Gray Matter and it became Treyarch.  Everyone that worked there got fat bonuses with the buyout except me.  Since I wasn't technically a Gray Matter employee.  So that sucked big time.  After it completed Gray Matter offer to hire me but instead I went to Raven Software with dreams of Hexen in my head.  Considering the insane success of Call of Duty I made another financially bad decision.

Anyway if you haven't played it go grab it. http://store.steampowered.com/app/9010/

Friday, October 7, 2016

Lotta Quake

Looking through my wardrobe I recall how many Quake games I've been a part of.  Certainly strange to thing that when I was playing Quake 1 I wasn't even part of the game industry.  Things certainly changed.


Monday, October 3, 2016

Sad News

Booted up the old PS2 tonight just for the hell of it.  Loaded War of Monsters and found one of the controllers no longer works.  Bummer :(


Saturday, October 1, 2016

Preparing for the Cold

The chill is happening out there here in Ontario.  In anticipation of the upcoming winter months I decided to clean out my art area a bit .  Believe it or not this is better than it was.  At least I can now actually see the desk.


Tuesday, September 13, 2016

Measuring My Art Chops

I've been tinkering around with a project at home.  It's a little overwhelming to sit down in front the computer and just create a game that isn't there in any form.  Where do you start?  What can you do yourself and what do you need help with?  These are the questions you don't have answers for until you dive in.

I have an art degree so I thought I might be able to handle the art myself but as a game designer it's been a long time since I created much art.  So I thought I should see what I can do in Photoshop.  My goal for the project's look is to have a style that was fun and sort of cartoonish, something like Warcraft.  To start I decided to do something simple and just create a board or a texture for a beam of wood.  No particular scale or metrics, just see if I could get the look I was going for.


This was sort of whimsical and fun like I wanted and was going the right direction but it needed more details like dings and chips.


So from start to finish I'd say it took about 3 hours and looked pretty close to what I had in mind.  At least I accomplished my goal of measuring my current art ability.  At the start I felt a little rusty but after turning on some music and digging in it started flowing pretty naturally.  I think it turned out alright.