Lee Cummings on pre-production

At the end of my childhood garden was a large arched wooden door with a heavy wrought iron handle framed by thick ivy. On the other side of the door was Steve Langley’s garden and house. A few doors down on the left from Steve’s house was Lee Cummings’ house. For much of 80’s this dog-leg route, through poor Jeff and Ingrid Langley’s home, was our Silk Road. Only the cargo was largely silicon not silk.

It went something like this. Out the back door, run down the side of
the Langley’s house, throw the ball to distract Tess the standard
poodle, quick wave to Ingrid at the sink with her marigolds, out onto
Hartland Road and down to Lee’s house – sometimes actually through the
Langley’s house if the dreaded padlock was on their side gate. Throw Raiders of the lost Ark on the 2600 to Lee, grab his copy of Star Raiders or Cosmic Ark, talk a bit about the dragons in Adventure and then back
the way I came. Plug in the cart, realise I forgot the additional
keypad and back out again, through the gate, throw, wave, run… And
so it was for Lee in reverse and through to the Commodore Amiga and
also a ton of VHS videos.

Fast forward some 25 years, Lee and I are still doing much the same,
only this time Jeff and Ingrid’s house is mercifully spared
thanks to technology like Steam, plus or minus a few dislocated hours
across transatlantic time zones. The constant is still our insane
obsession with video games. I honestly don’t think it will ever end, in fact now we’re both parents it will probably get worse!

Between 2002 and 2009 Lee Cummings worked on most of Rockstar’s
output, culminating in being one of the producers on San Andreas and
GTAIV. Here’s some of his “ramblings” on what that’s like.

“Milestone schedules are tricky things. It’s easy to sit back and say
“you know, it’s art, give it time / more time”, and that’s all well
and good until you have to answer to the guy holding the purse strings
for the project, and his boss, and then the marketing department about
why the date on all the banner ads and in all the previews is off by a
season or two, and trying to arrange dates for QA and ARGHHHHHHHHHHHHH
it’s slipped again, get me someone on the phone to scream at.

So we need to schedule. There is a chance on every project, with so
many variables outside of your control, and as with any situation
where you’re trying to make art (this applies to everybody on the
project, even if you’re a one man army something will do it’s best to
screw up even the best laid plans), one day you’ll end up in a
conversation with someone above your station who is liberally throwing
around majestic plurals focusing on the project needing to be under
control YESTERDAY.

Nobody likes schedules. I don’t think even people who spend all day,
every day, focusing on the minutia of game development schedules
actually like schedules as much as they like to say they do. Like
dreams, schedules are slippery little bastards which seem to make
complete and utter sense one minute, but quite quickly turn into crazy
ramblings the next when things slip, part of the project changes, and
nothing makes sense any more.

So nobody likes them, but we need them, so how do we work with them to
make them better?

Of course, the “we” in the statement above is laughable because
everyone who has any contact with a schedule in any way whatsoever
will have their own thoughts on how to, well, make it better. Creative
types will want a longer pre-production period to iterate, your
developer side producer may want the more “dangerous” parts of the
design designed and proven early, while the publisher side producer
has his own pressures which could well translate into having something
impressive and playable at certain dates for corporate / PR /
Marketing, and so on.

So for this “we” I’m talking about those who have an interest in
iterating as much as possible before things get too set in stone. I
assume it’s the same for bands – they’re sitting in the recording
studio just throwing out riffs which sound cool, while other parties
with their own vested interest get annoyed that nothing is actually
“getting done”, and suddenly something clicks, you have a number one
hit, and everyone has new hats made of money.

Miyamoto’s oft quoted “find the fun”, while limiting and arguably
inappropriate for lots of your development team, sums up
pre-production rather neatly. If you’re making anything at all
original, even if it’s a design which blatantly steals 90% of itself
from another game, or games, you need some time, however short, to get
that final 10% in some sort of shape before you start building the
rest of the game around it.

So here we are at day one. The deal is signed, congratulations–now
you have to actually make the game. How freaked out you are right now
depends on a lot of factors – remember that surreal concept thrown out
by someone during a greenlight meeting which you offhand agreed with
to shut them up? Yeah, that’s coming back to get you. The “yeah, we
can build that” discussions you had without talking to your
programmers first? Lets hope that nobody remembers those, right? No
matter who you are, as a designer on day one there is some part of
your brain screaming that you need more people, another is screaming
that you need more time, a healthy chunk is trying to decide where to
start, and the rest could well be pre-occupied by the fact that the
first milestone for the project is likely 4 – 6 weeks away and you now
have actual deliverables to produce on which real cash is tied.

Lets hope that the pre-production schedule is in some sort of
reasonable state to allow you to make the game the best it can be.

So, how long should pre-production be? yeah, we’re back to the “how
long is a piece of string” thing again, and based on a “full length”
project (2-3 years) where there is significant new untested design to
be brought into being I don’t think it’s at all out of order to aim
for around a third of the total development timeframe. That may sound
like a lot, but it gets eaten up quickly. Remember, you’re starting
from scratch so the first few months are going to focus on the
absolute basics — get your GDD, TDD and so on into a shape where you,
in detail, spell out what this game is going to be. You’ll be making a
multitude of incredibly important decisions, and while “how tall is
the player and how far can he leap” may sound boring at first, god
forbid you change your mind after your artists and level designers
have knocked out a few square miles of levels. And then you have the
back-end vertical slice to get out of the door, which will not only
have to prove out a chunk of your design, but will be a testing ground
for asset pipelines which may not exist yet, and will probably need a
full QA cycle.

Pre-production is a time of constant juggling, a time which will be
most productive if there is an amount of flexibility (if you get to
the end of the month where the weapons in your game are supposed to be
finished, but you’re sure that some extra time will make them
dramatically better, having the flexibility internally on the
developer side, and someone who can support you and run interference
on the publisher side as the milestone schedule changes, is an
incredibly important thing), but also a serious amount of discipline.
If you get to the end of pre-production and you’ve perfected the
weapons, but weapons are only 10% of your game, you’re probably
screwed.

So your schedule will need to take these things into account as best
it can. You’ll need fixed deliverables on each milestone, however
these should allow as much flexibility as possible. Month 3 may have
you working on general player locomotion, and at the end of month 4
you need animations X, Y and Z finalized. Or you’re working on
weapons, and the following month for weapons A, B and C you need the
models and effects finalized. Unless there is some very good reason
for naming those 3 in the milestone schedule (they’re risky on the
tech side, or have lots of dependencies, and so on), then don’t. “At
the end of month 4 we’ll have 3 weapons locked down” will give you the
freedom to iterate on parts of the game where, god forbid, you may
have found some serious fun which needs to be gently massaged out and
refined. If you find that the rocket launcher is changing the game in
ways you never expected, you won’t have to put that to one side and
rush to get the revolver fixed up just because some document written a
significant time ago is telling you to.

So iterate. Iterate often. While you’ll need parts of your team
focusing on parts of the game which need to be, as best they can,
“built once” or “changed in relatively minor ways through production”
(pipelines, low level network code, and so on), working on the
building blocks of gameplay you shouldn’t be aiming for final assets
early on in pre-production, you’re aiming for “final ideas” which get
built into final assets, or final assets are built around, later.

Your schedule should intelligently build and prove pieces of the game
which need to be in place for the vertical slice. Working backwards,
you need to define exactly what that vertical slice needs to show, as
a subset of your final proposed design. It shouldn’t need to show
absolutely everything, but core concepts, and your core gameplay loop
need to be proven out. From that vertical slice, you need to build
each component piece through the prior pre-production period. Start
with the high risk ones first. If you’ve promised that 4 player co-op
will be the backbone of your game, then you need to get any technical
issues out of the way as early as possible so that design (and
external expectations) have time to adjust properly if some huge
spanner gets thrown into the works. Try and keep the external
stakeholders in the project in the loop as much as is reasonable
throughout each individual milestone. Software such as confluence
makes this much easier than it was in the past, and having any
contentious issues brought up mid milestone when they can be tweaked
instead of 10 days after the milestone has been submitted means that
you’re not wasting time during the next milestone changing things
which could have been sorted out earlier. The importance of this
cannot be understated if you have a licensor in the mix — have them
sign off early and often on as much as possible so that you can build
upon decisions you’re pretty certain won’t change at a later date.

And finally, make sure that the everyone involved has their
expectations set as correctly as possible—that is, that they
understand that the primary goal of pre-production is not to just to
make something pretty. It might be pretty, but if it isn’t fun, if it
hasn’t managed to find that special glimmer of game play that would be
fun in a white box room with no textures on anything—than much time
has been wasted”

Lee C.



See and download the full gallery on posterous

Leave a Reply