|
[What can game developers learn from the film industry, if anything? No, it's not about storytelling -- it's about the very style of production, argues Tess Jones, who has worked as both a film producer and a game producer.]
Over the years I have mused on the differences and similarities between producing games and films. Both have large, creative crews working towards successful delivery of a visually entertaining product.
When I worked on movie sets, I drove around the city to a different location each day. Once there, I was greeted by a troupe of 200 creative people on the movie set all trying to achieve one vision.
When I worked on games, I was again greeted by 200 creative people all trying to achieve one vision, but instead of using a physical set to stage their dramatic scenes requiring me to cross town, the environments and sets were all contained at the office on their computer screens.
Despite their different work environments, both mediums aim to entertain, creating tension and excitement, making people laugh, cry, or tremble in fear at the edge of their seats.
From there, the similarities seem to end. Producing works in these two fields is drastically different. Films have significantly shorter production periods than games. A detailed schedule is created based on the scenes required in a screenplay. The cast and crew are hired, production begins, and each day they film specific scenes until the entire script is complete. When all scenes have been filmed, the crew is done. This can all be done in as short as a month.
Games have long production periods. New gameplay mechanics present engineering challenges. Players have the ability to stop and walk around in environments, rotating 360 degrees around objects. Unexpected bugs may arise late in production, not to mention the possibility that players will navigate levels in unexpected ways or become frustrated with gameplay elements requires ongoing iteration as testing happens. And finally, games are generally much longer than films, and require a hefty amount of creative content, with "short" games providing a six to eight hour game experience.
Despite these differences, I believe there are techniques from the film industry that can be applied to game production. Film production teams deliver fast because they have to, with location, crew, and cast restrictions tied to a very precise clock. As the market tightens and consumers expect more features from games, we need to find ways to make games faster and cheaper. One place to look is to the well-oiled machine of film production.
Lesson #1: Never Shoot a Movie without an Assistant Director
The cast arrives at 5am for make-up, while the production crew of 200 people gets there at 7. First up is a scene in a downtown office building, which includes a complicated crane shot. A second unit is shooting up the street to fill in the gaps so the whole crew can pack up and be at a second location by 2pm. The second location closes by 6pm -- no ifs, ands, or buts -- and they have to get four shots before the sun goes down, one including 50 extras in the scene. Oh, and by the way, your key actor is late, meaning you have to rearrange your entire shot list and pray to God you get everything complete without having to add another day to the schedule -- and budget.
Holy jigsaw puzzle of time management! If you thought your teams were hard to manage, imagine the pressure on the shoulders of a film's Assistant Director. "ADs," as they are known on set, are unionized through the Director's Guild of America.
They are highly skilled in judging all the various elements that will go into a shot and determining how much time it will take. On a film set where money is literally being spent as each minute on the clock ticks by, they keep things running smoothly towards completing each shot on the list.
I've worked on small films without an AD, and the inevitable result is that you find yourself still trying to "get that last shot" at 2am in an apartment in the Bronx, eventually falling asleep with your face plastered onto a piece of pizza. It's not pretty.
People tend to avoid the clock in games. Thinking about time estimates hampers the "cool" and "creative" game dev lifestyle. It's all about iteration, and you can't put a time estimate on that, can you? That's all well and fun during concept phase when your devs are passionate, but when you're exhausted and pushing to Beta... Yup -- you got it. You're stuck with another brutal, middle of the night sleeping pizza face incident. Sleep deprivation -- that is the real obstacle to creativity. What you need is a skilled AD.
What? "I don't need that! My producer does that." Well, yes and no. Some producers are amazing at time management, and others not so much. Producers often also have other elements on their mind: big picture concept, correspondence with marketing, milestone reports, a whole lot of other things that draw their attention away from the nitty-gritty, day to day of making sure elements are "in the can."
Movie sets have both a producer and AD, each managing different responsibilities. What game teams need is a dedicated resource to manage time. A qualified, experienced resource that can eyeball time estimates and build a schedule based on the risks and elements in front of them. Headcount is always tight on game teams, and project managers dedicated to scheduling could be seen as unnecessary overhead. But if you want to shoot a movie in 45 days with no overages and to have a beautiful film in the can, in the movie business, you hire a good AD.
|
To make that process more applicable to games, imagine instead of game designers and concept artists passing around "concept documents", game designer/programmers and concept/asset artists passing around game prototypes. The overall flow would still be preserved, and you'd have something concrete to show that has already been put through some of the iterative process of invention.
Similarly, with the printed game design bible idea, I think your thought of regularly playing the game and that itself being the point of reference is more on the money.
Thanks for the write-up, this is an interesting and thought provoking article :)
There is a TV show on the Biography channel called "Inside Story" that examines the making of great films by interviewing the director, actors, etc. It's amazing to see how much iteration and experimentation happen with these movies after the script is written.
Game development is more like science: hypothesis, preliminary research, reevaluation, follow-up research, potential future research... with the next step unknown until the end of each stage. You're always building on existing research, but any time you delve into new territory there's no telling how it will play out.
Maybe we need a peer review process instead.
Personally I'd say people shouldn't overlook the importance of preproduction. I'm not going to say you need the "game bible" from day 1 and describe exactly what the game is about, but having a solid concept is important. If not, what would prevent devs from slapping a 10-head mutalt soldier into WarShooter 9,000 because the iteration process proved it was cool? I know this is an exaggeration but you get the point.
I think this article can help us rethink many things and maybe find a more cost-effective ways to make things. After all, even if this is an entertainment industry, things still revolve around money and anything that can help us invest less in games is very welcome, specially considering the current economy.
You also often hear (at GDC, etc.) that if management gets early buy-in in a concept, it doesn't have to get reworked during full production, as often happens when licensors and/or publisher management see a game that's well into dev and hate what it's all about.
So yeah -- I would argue that for these two reasons it's not so easy to dismiss.
Jason, I think what you are saying is definitely on point. It would make the most sense if the concepts weren’t just documents being passed around, but were an actual, small but playable example of the game. That said, if you had an army of people making game concepts across the world and submitting them to different game studios, I can’t see how that approach wouldn’t work in a very practical sense, since every game studio has different technology. Maybe there is something in between these two ends of the spectrum. Perhaps the concepts would include a package of art assets including game character mockups, environment concepts, storyline, and a suggested list of missions and gameplay elements. The package could primarily be art and story/concept driven, but the game team would still need to come up with gameplay foundation to fit that art/story package to fit with whatever type, size, and platform for the game they want to make. But at least the teams would have the ability to sift through hundreds of story and character ideas that they could then purchase and augment as they see fit. The forethought that would go into these packages, and the volume of the number of high-quality concepts that studios could choose from, would be the win.
I do have to say that the point in the article about concepting was mostly theoretical, as it would require such a huge paradigm shift in games that it is most likely impractical to consider. But it’s fun to discuss!
Easiest way to make it happen seems to be for a large studio to take sub-groups of developers and let them explore wild new high-risk concepts with prototypes and basic art. You end up with a pile of wild ideas based on internal tech. Most won't work, but there might be a diamond there.
Now that I think on it, I believe we've seen this sort of thing work very well. Valve did it, as did Double Fine, both with strong results.
@Jamie - I also have to take a little exception to the statement that in both filmmaking and cooking "Prepare the ingredients and follow a basic set of directions. The strength is in the recipe, not the process". Both recipe (script) and process are of equal importance to a great outcome. Yes, you can't make a great film without a great script. But bad, or even middling, execution of an otherwise flawless script (or recipe) will simply kill the original vision.
To complete your cooking analogy, you should probably should analogize game development to baking... Even among cooks, baking is known as the more "scientific" of the culinary disciplines. Baking is a lot less forgiving, and permits a lot less improvisation than other cooking.
Waterfall development is going by the wayside. Game design doesn't work like that. You're not operating from a premade recipe. You start with a recipe for something you've made before, and then you switch it up. Game dev branches out into new areas of knowledge, from technology to story interactivity and player psychology. It learns from past mistakes (ideally) and attempts new things (ideally). TES: Oblivion experimented with the "Radiant AI" system; Skyrim builds on that system and tries out many other new things.
Part of the reason my analogy is cooking versus science is that the two production processes are *not* variants of the same discipline. They have fundamental differences in purpose and methodology. It's true that some lessons can cross over, but most do not.
Game development can and should learn from every other form of media, and every field of knowledge: books, music, TV, economics, psychology, everything.The best game developers often know a lot about fields unrelated to games themselves, because that is where we find growth as a medium.
TL;DR: Electronic games are all about crossover and integration of seemingly unrelated fields, math, science, and art. Film production is about writing a story and then telling it in a visual form. They both can tell a story, they both employ technology to do it, but there is a fundamental difference in purpose and scope.
In general, for the concept idea, I was in no way suggesting that we should make games from a recipe. I was mostly just theorizing a different model for concepting art and story for games, since games are getting bigger. This would not work for casual games where the concept is often the gameplay itself, but would be for bigger console titles that employ characters, big environments, and stories that drive player attachment to the product. The concept packages would have definitely limitations and I can’t imagine that they would include much in the way of gameplay elements because of technical limitations. The teams themselves would still need to create fun gameplay and face their own technical challenges head-on.
Dominic, as for script development, I definitely agree that sometimes by the end it is “script by committee.” I imagine that could be a good or a bad thing, depending on the original script and the crew involved. For what is applicable to games from the screenplay process, I was more interested in the script purchase process, before it goes into development with a director or team. The idea that all these stories are floating around and studios can pick which ones they like and then purchase them, is where films differ from games. Even after script development and production on movies, I would think that having a strong starting point at least reduces the chances of failure on movies as a whole.
Maybe the games industry needs an equivalent short form to spur creativity. Indie games don't exactly fit the criteria, because they're more removed from commercial pressure than commercials and music videos. Maybe mainstream downloadables and DLC could fit the bill. Perhaps the industry could benefit from consciously adopting that approach. I don't know all that much about advertising, so I could be off here, but it's just a thought.
That's a great point. And having "lots of resources" can be almost as bad as unlimited resources. You can easily end up wasting those resources, because you end up trying to make "everything" really great, and you end up spending a bunch of time on a difficult and complex feature which has only a very minor impact on the player's experience.
The thing I'm working on right now is the poster-child for this: its a minor aesthetic feature with no gameplay consequences whatsoever, and yet its sucked up several days of time from a programmer (me) and involves also modelling, animation, mocap. Its fun to work on, but its an expensive feature considering that it has no effect on the gameplay. This is for a big project though, on a more modestly-sized project with limited resources, this feature would have been rejected after about 30 seconds of discussion.
The problem with unlimited resources is that someone in charge can say "I want a pony" and three weeks and a small fortune will then be spent building one!
#4 and #7 (crunch time and food) are important points, since they affect the morale of the team more directly than the others. If you can't spend time with your family, at least you get to eat a steak. Not that those items compare, but morale wins battles. The powers that be should do everything they can to keep it up, and great food works wonders. And overtime pay may be a sensitive topic, but it does place the responsibility on the right shoulders and, like was said, keep morale up. Once, an otherwise dreadful 20 hour day became nothing short of awesome because meal penalties, combined with overtime, really shines on a paycheck.
Regarding post-production, since the elements film associates with that time are developed in tandem with the rest of the game, leaving sound et. all for "later" doesn't work in game development. Similarly, I'd agree the script vetting process would need major overhauls for game development.
But, no one would argue these industries are the same (I hope). From these differences we can still learn a lot and adapt. Film production has what, almost a century on us? Of course, we always talk about film production, but I wonder from which other industries might game development benefit in surprising ways.
When devs do things like change the story at the end of the project (Mirror's Edge), it's like you've rendered a beautiful drawing in graphite, and then decide to change the pose. You can do it, but it will be a pain in the ass. You should've just worked harder when you were drawing gestures.
In that sense I think the first part of the article has the right idea, but gets mired in details that don't translate to the game industry.
The idea that game development tasks aren't carefully and precisely time-estimated just because it's not 'cool' is so completely, fundamentally wrong-headed that I can't even begin to imagine what kind of mindset would lead to it. Time-boxing engineering tasks is tremendously, astoundingly difficult and no software engineering discipline in any industry that I am aware of has come up with a reliable way of doing it. When you combine the difficulty of estimating time for software engineering tasks with a creative endeavor like game design I just can't imagine how you can ever get anything close to accurate estimates - I've certainly never seen it done. Once you accept that your estimates aren't going to be accurate, they become much less useful as a scheduling tool and it can make more sense to ditch them entirely, depending on how your product is being built and what your milestones look like.
And to be clear, I say this as someone with real experience as a designer, programmer, and producer - so I feel like I at least have a little perspective on this. Maybe that's why I found the author's tone so offensive on this subject.
But I guess that's also the nature of game development as well, and as we mature as an industry I'm sure that we'd find a collaborative way to solve this issue.
Great Article
Maybe a better analogy is the Japanese media system. It's integrated in a chaotic but fascinating and I believe powerful way. Consider manga the source of concepts (it isn't always), top manga is determined somewhat by editors, but more often by actual audiences. So much new manga is printed every week, the cream of the crop really is the cream of the crop by the time it's picked to become anime. Then, those anime which are wildly successful, (again due to audiences) are adapted into video games. Clearly, that's an oversimplification, but it is a common case. Great ideas, especially unique ones, rise to the top far more often than in American.
In America, our culture limits this wonderful mechanism. New comic books are targeted at niche markets. Regular print fiction is read by a handful of the population. Those potential sources for popular game concepts is sadly weak in the USA.
I'm afraid it's also extremely misled and misleading. It's anti-advice. You would be better off never having read it. Especially if you haven't worked in game development for a while.
For decades, games production has suffered horribly from the blind importing of film production methods. Yes, both media go on screens. That's about the end of the similarity. There are fundamental structural differences in the product which invalidate the assumptions underlying the film production process.
Just to start with the assistant director part:
ADs can only do their job because they know how long things take. After years of experience they've basically figured out how long it takes to get extras together, set up a certain kind of camera crane, light a scene of a certain complexity, and so on. The reason is because these are primarily repeatable, physical tasks. You're not inventing anything new. You're just getting something done. So you can estimate to a good accuracy how long they will take. And since you know that, you can shuffle tasks to line up as needed.
In games, nearly every task, at nearly every stage of production, is inventive, not physical. It's easy to type code into a computer or click a mouse on an art program. But we don't schedule the physical of typing code or the physical process of mouse clicks. We schedule the creative process of engine design or artistic development. Creative, inventive tasks are fundamentally unpredictable because they deal with significant unknowns. Since these tasks are so unpredictable, times can't be predicted nearly as accurately as with a task like assembling extras on set. Which means that the whole "skilled manager shuffling tasks around like cards in a deck so they line up perfectly" thing is a dangerous mirage. It doesn't work. Nobody can do that in game dev. If you try, you suffer (and crunch). Many have. Many still do.
In games, we need to design the process to handle the extreme degree of uncertainty attached to every plan and estimate. Hence iteration which, while it has its downsides, is antifragile against the kinds of unpredictable outcomes that regularly happen in game dev.
Script development:
Again, the prediction problem. A film, even if it's bad, will do what it was told to do in the script. The good guy will always say the same line and kill the bad guy the same way. It is structurally inviolate. The experience might suck, but you know what the experience will be.
Games are interactive. You don't know how players will react. Nobody can because players are unpredictable. So you can't know the experience that is being created until it's tested. So you can't plan a game on paper like you plan a movie. It will disintegrate on the first playtest. You need playtests to even understand what experience you're creating at all. And to playtest, you need a working game.
Think of it this way: Reading the script and imagining the film in your head is akin to playing a prototype and imagining what it will look like when fully polished with art. You _cannot_ go from a written design document directly to a coherent and correct mental image of play.
Again, the solution is iteration.
Story equals concept:
This is a worthwhile notion to keep in mind. But don't apply it too blindly to games.
Game stories work best when they integrate with the gameplay instead of running alongside it. See Portal for a great example. This means that as the gameplay shifts during iteration (which as noted above is essential), the story must shift to match it.
So, I think we'd be well served to try to find high concepts over the course of development, but not in a big lump up front.
I don't really want to go on. Some of the later advice fits much better with games than what's above. But my general conclusion remains: be very wary of borrowed methods from film. There are hidden problems in them which spring from assumptions which hold in film production but not in games production. We already borrow too much from that medium without realizing it.
We need to develop methods native to our medium, not steal even more voraciously.
Film production is never a straight line process. All through the writing, production, and post-production of a film you are trying different things, hitting roadblocks, finding stuff that doesn't work for any number of reasons and trying to change it while you still can.
The actual production of a film is chaotic and messy. Any number of roadblocks can come in and require radical reinvention of what was intended to be done for the story. It's high stakes, high pressure, and requires a need to be incredibly flexible with how the source material is approached. There's a reason that people in the film business call those outside the business "civilians".
The post-production process also, particularly editing, is highly iterative. When you've finished shooting, you have all this footage you have then turn into a cohesive narrative. Editors and directors try out numerous options before settling on a final cut. Individual frames can make a huge difference in the quality of a narrative.
I suspect that even many of the borrowed methods we take from film aren't actually the methods used in real filmmaking - just a naive outsider's assumption of what they are. Which is even worse.
But just based on Tess' article, it sounds pretty darn non-iterative. A lot of waterfall-style stuff there.
Follow their documentaries and making-of features, you'll see them describe why they'll make a full storyboard animatic (blueprint) and then previs (prototype) of the entire film before spending money staffing up production. Heh, reason being that they do it because they want to make all the potentially expensive changes BEFORE production - so they don't lose any significant chunks of time or money that could be spent on polish and making a superior film.
I started researching them (and companies like Valve and Halfbrick) directly back in 2008 because I had started getting involved in the management side of things and couldn't find any books or resources on production management in animation and games.
Heh. There still aren't many good books or resources on it :P.
I've been applying methods from tech startup guides like ReWork and The Lean Startup and converting what's worked with my teams into articles and guides on www.indiebits.com
ReWork: http://www.amazon.com/Rework-Jason-Fried/dp/0307463745
The Lean Startup: http://www.amazon.com/The-Lean-Startup-Entrepreneurs-Continuous/dp/030788 7898
Hope that helps!
I'm aware of the storyboarding process. I'm confused why you would mention this as a kind of iteration. This is the opposite of iteration. It's planning up front and then executing the plan exactly, piece by piece. Like blueprinting a building and then getting workers to put it up. Not iterative.
Obviously they iterate a ton on the storyboard/script itself. But from what you're saying and from what else I've seen, once you're in production it typically comes out looking just about like the storyboard does.
Try to do that with games, and you're dead meat. You can't storyboard or write a game down and have it work. You have to build the game and playtest it to know how players will respond. So it changes all the way through production, at every level. Even after the basic prototyping is done.
Being native or not native has nothing to do with it. The only reasons these suggestions would not work is if you have poor management or poor creative vision. Certainly that is in an abundance, but that is no reason to dismiss the ideas blindly.
It's great advice that most companies in most industries would benefit from, but which, unfortunately, is so tough to measure that corporations are often likely to dismiss it. It takes progressive or "hip" companies like Google, Best Buy, Microsoft, to throw in these kinds of "perks", which actually end up being perks for the company via not just morale but employee psychological synergies.
Yeah, the rest of these points could reasonably be debated, but that one is golden. It doesn't have to be all the time or any kinds of lavish, but the act and the effort can pay significant dividends.
Just take that in for a moment. How many of you spent months (if not years!) on a chunk of the game that never made it into the release build? How much money do you think the studio spent producing that wasted material?
Games could save millions on production by taking the same approach as film does with previs by creating a horizontal slice of the entire user experience during pre-production. Yes you'll be making plenty of changes during development - but the big cuts can happen before they cost you time and money.
I worked in film for 5 years before getting back into the games industry. Yes there's a lot that would benefit games (most of which is brilliantly described in this article) but the film industry itself is also deeply flawed.
The entire reason I got into business in the first place was because every animated film production I worked on ended up burning people out, going over budget and either a) destroying the studio or b) leaving the studio with a skeleton crew struggling after another project.
#1 thing film needs to learn from games: Pipeline Optimization.
Happy Feet could have saved a few million dollars just on improving render time. And having worked on 85% of the shots in that first film I can confidently tell you that their use of polygons and textures was grossly wasteful.
Your notes on horizontal slices sounds like an argument for broad graybox prototyping across the whole project.
The indies startups I coach can't actually support themselves long enough to do the "make vertical slice, pitch to publishers, pray for funding" model.
The model we follow is 100% about getting enough cash-flow during each stage of production to keep the lights on until the game is ready to publish. It looks like: Brainstorm > Test Concept > Horizontal slice > Monetize > Alpha > Monetize > Beta > Monetize > Gold > Monetize > DLC > etc.
Far prefer this method to developing in isolation and hoping and praying that people like it.
But actually, repeated market contact with minimum viable product isn't a bad or risk-adding thing. In fact it's probably quite risk-reducing.
The difference of course is that you have seen this from the inside, whereas I've very little insider knowledge from any of these industries. I know there are some authors on this site. Has anyone worked in TV?
Trying to shoe-horn one production methodology onto another is like a bad analogy; it will make sense in a couple of situations (maybe only one) but it won't work for everything. Streamlining production is about analyzing YOUR situation and adjusting accordingly...it is not about superimposing another methodology on your situation and forcing a fit.
"I have never met a man so ignorant that I couldn't learn something from him."
- Galileo Galilei
I'm sure a tug boat captain could learn something from a water skier. Similarly, if anything, you learned what not to try from this article. Take from it what you will, and you've still managed to take something.