Blue Sky Thinking
With Project Skyline, Autodesk's Marc Stevens and Eric Plante are bringing the disparate worlds of art and programming together
As game budgets continue to rise, the need for efficient use of time and resources becomes ever more important. However, the traditional production pipeline is fundamentally flawed, the artists surrendering their assets to programmers with no effective way of spotting mistakes and make adjustments on the fly.
Only companies with enough resources or technical knowledge to create proprietary solutions have been able to address the problem, but Autodesk is here to help. With Project Skyline, it is working on a new model for game production that creates a live link between the art assets, the game editor and the run-time, unifying the entire pipeline.
In this interview, Autodesk's vice president of games Marc Stevens and character animation product manager Eric Plante walk us through Project Skyline, and its potential repercussions for the industry.
Q: Autodesk has been steadily ramping up its involvement in the games industry. How pervasive are you products now? How much opportunity is there for the future?
Marc Stevens: No matter which successful games you look at today, they're using one or many of our tools to create the content. Looking back to a few years ago, there was a lot of money that people were investing, studio-by-studio, project-by-project. It was all their special sauce but, when you got down into it, it was all the same stuff, and it's because you couldn't go anywhere and get a solution.
So everything that the developers did - the programmers, the game designers - most of that was in-house developed stuff. Even a couple of years later, even the most successful third-party tools for high-end titles Epic or something they sell 25 or 30 licenses a year, so you're still only touching a handful of games out of hundreds that are made.
And then you've got this boom of all the mobile games that are being done, and all the social games. Yes, they're a lot simpler than console games, but on the other side mobile hardware is getting more powerful every year. The iPad or the smartphone is going to be just as powerful as a PC or a console, so there's no reason why the content or the complexity of the games can't go with it.
Q: What does Skyline add to what you already offer?
Marc Stevens: One of the areas that we thought we could get involved and help is to bridge the gap between what all the artists are doing, and what happens on the other side of the pipeline. We could see there's a lot of inefficiency, and energy and time lost.
The smartphone is going to be just as powerful as a PC or a console, so there's no reason why the content can't go with it.
Marc Stevens, Autodesk
That information [art assets], usually it's all transformed once into a level-editing environment that's built specifically for the game to lay out, design the gameplay, and all of that stuff and when it's time to make the final game, it's converted again. And whenever something's wrong you want to make changes, there's bugs finding out what happened and why and then changing it is very hard, time consuming, and complicated.
We've seen some of the more innovative companies designing processes and doing things on their own to solve this, but Autodesk is in a position to help them, too. If you roll the clock back a couple of years that's one of the reasons why Autodesk got in the middleware business: to learn more about the problems the programmers and the developers are having what happens in the run-time because it was an area we just weren't that familiar with.
We started with the acquisition of Kynapse, an AI path-finding middleware. From there we did some of our own stuff with Human IK, procedural animation and motion-retargeting middleware. Then [we acquired] Beast from Illuminate Labs, for lighting, and most recently Scaleform for UI.
Q: How many people at Autodesk have direct experience of making games?
Marc Stevens: Eric actually came from EA, here in Montreal, he worked there for a number of years. And Mathieu Mazerolle, who couldn't ne here today, came from Ubisoft. Another guy was Frank DeLise, who worked with THQ on the Homefront franchise and a bunch of other things.
I don't want to say that we didn't have people here that knew about games, but I don't think we had people who had spent ten years making games... We started to bring in people who had to go through all of this; that used our tools and saw the inefficiencies, so we could start to look at how to build a more unified process.
Q: Can you explain what you mean by "unified process"?
Marc Stevens: Where the artist can both create content that can be merged into a run-time authoring environment, and also be brought out to the game in a consistent manner where people can make changes and see the results right away, and if there are problems understand how to backtrack to where they came from in our tools. That's where we started from with Project Skyline.
Q: Project Skyline offers a unified pipeline specifically for character animation. Why is that?
Marc Stevens: Saying you're going to provide everyone with everything they're going to need to make a game is a big, bold statement. I don't think we were ready to be that bold... The character animation problem touches a lot of things: there's a lot of content creation on the tools side, there's a lot of run-time authoring, and then there's run-time; it kind of touched all of the pieces so it was a complex process. It wasn't an easy problem.
It was also a problem that our customers identified as difficult for them from a workflow perspective. So that's why we decided to take a vertical slice.
Q: But the basic concept could be applied to other areas of the production pipeline?
Marc Stevens: Skyline is a platform for us to take this model to whatever, whether it's lighting, layout and design of levels. We took character animation as a vertical slice, so we could go from beginning to end and at least validate that what we're thinking about in terms of solving this problem makes sense, and it's actually going to help the customers. Once we do that, we can then apply this approach to other areas of game design.
Q: How important is this idea of a unified pipeline to game development?
Eric Plante: The economics of making games has changed over the last ten years. When you're trying to make a big success, it used to be that you didn't worry about cost; you were just trying to make the best possible game, whatever the cost, and you'd make money in the end. That's just not the case anymore. Game makers have to become smarter and more efficient.
When I started at EA, things were almost always as I'm about to describe them. There was a pipeline where you start in a tool like 3DS Max or Maya typically an Autodesk product and you author your source assets there. Then you'd export those assets out to somewhere you assemble them to gameplay.
Only the studios with more resources are able to do this, and they do it because they have no choice
Eric Plante, Autodesk
First off, there are a lot of issues with that step. If you were going to design something from the ground up, you'd never design it that way. That would be one and the same. Once you've crossed over there's no going back. You can't make changes in the gameplay editor to your source assets.
And then you build those assets together and build the data for the run-time. And in addition to the same objection that this step also should not exist, it's extraordinarily slow. If you're unlucky, you're an artist, and your change depends on a programmers change... You look at the result of your work only once you've gone through all of this, and then it's never perfect the first time. You go through it again, and it's still not right - a bit better, but... So you go through that cycle many times and you lose a lot of time. It can be hours or days or weeks depending, for every one of those iterations. This is how we were developing games on my projects, and it's just terrible.
The more technologically minded studios implemented something like this at least for the bigger pain points in their productions maybe not across the board, but where they suffered most they invested in developing [solutions].
Marc Stevens: What we've learned from working with creative people in the film industry, too things get better with iteration. You don't get anyone going, "Right, that's it. This is what I want. I'm done." So when you look at that process the more iteration you can do the better your end product is going to be.
Q: So Skyline will allows the developer to iterate and prototype more during production.
Marc Stevens: Yeah. Absolutely.
Q: How does the process work?
Eric Plante: Skyline has an asset database that allows you to, when you see something on the run-time, it allows you to make the link back to the source asset... Normally, the authoring tool is outside of that live loop, so you're able to change how the assets are assembled together but you can't change the assets themselves, but when you think of it, that's the part that would benefit the most from being in that loop.
So that's one issue. The other issue Marc has touched on already: people are developing this on their own and it's very expensive to do. Only the more technologically minded studios and the studios with more resources are able to do this, and they do it because they have no choice, but really it doesn't add value to the game... That's what Skyline is about. We didn't invent this, but we're trying to standardise it, and more importantly, we also put the authoring into it.
Q: Has anyone tried to do this in the past? Unify the authoring tools with the editor and the run-time?
Marc Stevens: Not from a commercial point-of-view. Game companies had built up a lot of internal knowledge over 20 years, where they've made some things that look like this, maybe tied to a specific game, but also where the authoring tools aren't really well integrated into the process. So from the level editor out to the game, that part of the loop is probably a lot better, but including all the way back to the 3DS Max or Maya [asset], you don't really see that, even internally.
Eric Plante: We've done this for one vertical in particular: character animation. That's for a few reasons: one, it's for one of our core competencies; two, animation is an area that really benefits tremendously from this. Animators are trained to replicate and stylise the natural motions of the human body, but we ask them to author tiny clips of a walk cycle that's looping, of one jump, but it's hard to author those in the context of the whole motion of the character.
If they were able to author these animations in the context of the all the animations that already exist, it's a huge gain for them, and a completely different way of working. So we do this to allow animations to be authored, assembled together, and viewed all at once, live, as the game is running.
Q: From the demonstration, it looks very simple to make changes to almost anything about the character.
Eric Plante: It looks very natural, and just the way it should be...but it turns out it this actually just saved a week of somebody's time.
Q: It also seemed to allow a greater degree of experimentation.
Eric Plante: Yeah, absolutely. You can add, for example, a cover behaviour, or a vault behaviour, and just see how it works in the game.
Marc Stevens: It's not just a story of we save you money. It's also about giving people more time to experiment with ideas, and get a better end product.
Q: You showed this for the first time at GDC this year. What sort of response did it get from the developers there?
Marc Stevens: At a high level people seem very happy to see this. Everyone says, "Yep, this is exactly what we need. When can we have it?" I think the rubber hits the road when you have to start integrating this into your engine and your whole process, and where we're at now with this is starting to work with clients, doing initial integrations, so we'll learn a lot from that.
Eric Plante: People are extremely excited. It predates GDC, actually. When we work on things like this we work very closely with customers. We don't develop in a vacuum, and as I said earlier, we didn't just invent all of this. It's stuff that people have been doing at great cost, unfortunately internally, to some degree or another.
The big challenge for us is that people like the concept; it's just that, will they afford themselves the time to change
Marc Stevens, Autodesk
Marc Stevens: The big challenge for us is that people like the concept, they're open to it; it's just that, will they afford themselves the time to change.
Q: Do you expect the companies with internal solutions to adopt this, or is this really for people who have been forced to work inefficiently through a lack of resources or technical knowledge?
Marc Stevens: The latter for sure. I think the former it's going to depend on...If you look at a lot of the larger companies, they're struggling right now to keep costs down and keep these big teams going. We have talked to some of them and it's a case of, 'If you have a solution for me and I don't have to do it, I'm happy to do that.' At a high level, conceptually, they know they make their money off being creative and creating new, interesting games, not necessarily rewriting technology all the time.
I don't know if you saw the Carmack quote that came out, where he talked about, for the first time, we're not going to be replacing all of the technology, we're going to focus more on the creative aspects and that side of things. I would say that's where film-making is more ahead right now, just because the technology hasn't been there in games, and every new hardware cycle means a change of everything for everyone. It's been expensive. It's been tough. But it's starting to balance itself out now, and people can afford to standardise a bit more on the pipeline.
Everything's maturing enough now that we should be able to do better at this, and I think Skyline shows that we can. If we can get some good projects and good customers on board to show more proof of that we'll have a good story.
Q: Will larger companies want to keep the idea of standardisation at arm's length. If it allows developers with fewer resources to do more, and the democratic affect that could have, it might not be in their interests.
Marc Stevens: If you look at it that way, it ends up... Take film, for example, once things standardise the big studios only get bigger. Who makes the big CG movies now? Pixar, Dreamworks, Sony, Blue Sky maybe, and that's about it. You've got four or five that can really do it to the right quality level, and to really go high-end in games will still be a big undertaking, I think. Not anyone in their garage with a good tool will be able to pick up and do that. There's just a scale and magnitude that you can't match.
Eric Plante: Regardless of the sophistication level or the amount of investment that a company has in their own tools, everyone has technology cycles. There's always a point when you need to look at some part of your pipeline that has aged, and that's the point where you have to ask yourself, are we gonna do another one of these, or are going to use this solution that's now on the market that's fine for everyone else.
Q: So the standardisation of technology and development tools is going to be widespread?
Marc Stevens: I think it's going to happen; it's just a question of when. If you look at other industries, everything standardises. It's just a matter of time. Are we at the right time here? I hope so.