Art Direction Bootcamp: An expert roundtable Q&A 2

Art direction is not just something art directors do. Art direction is in every brush stroke we lay down and in every vertex we move. It is the idea that you're trying to communicate and the emotion you put into it. It is the dialog that you're having with your audience, while the tools are just means for delivery. 

If you care about art, we would be privileged to see you join us in San Francisco on February 27 for an entire day of deep and thought-provoking art discussion from the absolute best in the business, during GDC's Art Direction Bootcamp. If you're going to Game Developers Conference -- don't miss it!

In the meantime, We've gathered some of the most experienced and talented artists in the game industry to answer a few questions from Twitter and around the web:

Let's begin!

- Andrew Maximov asked:

How do you feel about more procedural and scanned content making its way into game production? Do you feel like it impedes or improves your ability to effectively communicate artistic vision and how would you like to see it evolve in the coming years?

Greg: One of the biggest problems we face as artists is the continual need to increase the content provided in our games. Any technology that allows us to increase content can be hugely valuable.  Something like photogrammetry is a process that could allow us to add content quickly for secondary and tertiary elements, so we can move on to more important content.  Ultimately, these are just tools and it is up to us as artist to find appropriate ways to use them and innovate with them in more artistic ways.  I think these content options are going to find their way into our pipelines and can’t wait to see what happens when artists really start to work with this kind of technology on a larger scale.

Brian: I'm a huge an advocate of scanning, and I look forward to using procedural content in the future to expand the size of our game worlds with less labor. I started working with photogrammetry on Tomb Raider. The goal was to create realistic proportions, fold logic, and deformation.  We captured the data in neutral and extreme poses and used dynamic blend shapes to drive muscle and fabric deformation.

http://tombraider.tumblr.com/post/131036893850/dev-blog-building-from-life-kam-yu-lead

https://www.youtube.com/watch?v=IUA70GncFsg

When I joined Infinity Ward I worked closely with Bernardo Antoniazzi who developed an amazing in house Photogrammetry system to create photo realistic faces. This process allowed us to capture likeness and facial deformations that translate into a believable performance.  Below are some links of the original R&D tests and pre-rendered and real time cinematics we created for Infinite Warfare using this tech.

https://www.youtube.com/watch?v=e8EhWPJYHMM

https://www.youtube.com/watch?v=y_RI3bZhU50&t=12s

https://www.youtube.com/watch?v=5kNLrj1qQaQ

Call of Duty has always focused on creating authentic worlds and our early tests expanding the use of Photogrammetry for environments, props and materials is very promising. As we learn more about PBR, scanning has allowed us to capture accurate surface detail, albedo (Color) , specular information to ensure every asset in the game is balanced and cohesive.  

To artists out there concerned about scanning limiting creativity, there will always be art direction that doesn't require realism and hand crafted assets will still be required. If you make realistic looking games, I believe this technology will allow us to put our artistic energy into creating worlds and experiences instead of assets. 

Denis: Procedural tools are just amazing. I feel for games that aim for realism these tools are already must haves. Clearly they still need some time to break in stylized art but I feel its something that is inevitably to happen. Iteration and content creation is what makes games expensive. The more tools will accommodate for this the smoother development will be. Which translates directly into more polished games and more content

Jason: As someone who loves to light, I am excited by this new tech and look forward to seeing it evolve over the next year or so. Not having spent a significant amount of time in this limits first hand experience, but the small amount thus far is encouraging. With any new software, tech or approach comes the learning curve.  This is accompanied by a rigid physically based boundary that can be hard to break out of when trying to explore. 

The long term challenge: Can we use that perfect scanned data simply as a base and springboard to a variant of that texture, while still maintaining all the physical based accuracy and detail from the scan?  On top of all that...can we get the varied look that our games need?  

- Twitter user Ara Carrasco asked:

What's the most important aspect of art direction for you?good lighting, good modelling & texturing...?

Jason: Practically speaking, I usually say lighting as it can alter the mood/tone of a scene with a simple dawning of the light.  Theoretically speaking, I would say tonal cohesion between what the game/film/movie is trying to represent with what the art is trying to communicate.  

Andrew: I'd have to say that it's the name for me. "Direction". I really like the word. I think we forget what it means sometimes behind the semantics of a job description, but directing for me is about asking the right questions rather the giving the right answers. I'm constantly amazed by how inventively creative people can solve a problem that you put in front of them. And as long as you are providing them with the right bounds and goals you get something that evolves way beyond what you could have initially anticipated. Power amplification is the name of the game. It sure is a lot of work to prevent things from going in diametrically opposite directions but to me it feels like if the artists have the room to roam within the bounds of the fictional worlds that they are creating the vision and direction get amplified tenfold providing for really rich and exciting experiences.

Greg: I would have to say that the most important part of Art Direction is the presentation.  Establishing where the camera is and how the game is going to be experienced by the player is paramount to me.  This is where the picture as a whole comes into focus.  It allows you to concept and design elements at an appropriate detail level.  In a lot of ways figuring out the presentation gives you the edges of your canvas.  It is the boundary that you create the picture in.  From there everything else flows.

Brian: Answer: I always start with the big picture, for environments it's Composition, Lighting, Motion and Context (What is the story of the environment) For characters, Silhouette and Motion.

Denis: That's a tough one there is so much to it. If I'd had to pick, I probably go with good communication and keeping your team motivated. Communication is key to optimize and limit iteration cycles. For that to happen it has to be a two sided dialogue, so you as AD can take in feedback and act according to it. For me personally the artistic decision making often happens during discussions with my team and I just make sure I follow through with it and keep people on track.

Motivation is essential for me because it keeps the team engaged and they will push for the best possible result, sometimes way beyond what was initially asked of them.

Both of these points in the end have direct impact on productivity, quality and efficiency - hey I'm German after all ;P 

- Twitter user Evan asked:

How do you balance advanced and detail visuals supported by modern consoles with practical design that supports gameplay? As in, keep clear indicators for the players while looking natural and not contrived for gameplay purposes at the same time?

Brian: It's always a conversation about goals.  On Tomb Raider we always tried to create believable worlds that the player instinctivly understood what was interactive or where they were supposed to go.  We use "Visual Language" to help guide the player with consistent visual cues. We mark climbable ledges with white paint, rope coils are always used to indicate where a rope arrow can go, bright pocked walls indicated axe climb surfaces and specifically cracked walls indicated places you could open with your axe.  We protect these patterns in the game so the player always knows how to interact when these examples of visual language show up in the game.

Greg: In a game like XCOM there are many design requirements that effect the art.  It is beneficial to try and identify what needs to be communicated by design early so that visual opportunities are more easily identified.  Being a turned based game can be seen as a huge challenge to immersion, but we saw this as one of our opportunities.  The camera in XCOM is a great example of this since it can literally be anywhere while you are playing.  By changing the camera we can reinforce the design mechanics and provide a tension that is difficult to convey from a more three quarter view.  However, the camera changes also provide some real art challenges.  We needed to make assets that read clearly from the top down view as well as in a close up cinematic camera.  This led us to the initial art direction of XCOM: Enemy Unknown that focused on a “miniatures” look with slightly exaggerated forms and more chunky geometry.  We didn’t want have the details be so small for the close up cameras that they would dance or shimmer from the default gameplay camera.  On the flip side we couldn’t make things so chunky that they destroyed the immersion when the camera dropped down. We spent a lot of time initially balancing these issues and setting very clear modeling and texturing rules.

Andrew:  Design dependencies are always a tricky subject. Confused or frustrated players can't enjoy the game so keeping them happy and their goals clear is a big part of art production. For us it's a constant back and forth. At the very beginning of a project we would iterate with design on gameplay relevant art elements to make sure we can produce art around it that integrates with it nicely. We always want to make sure not to muddy up our language. Exceptions are dangerous and confusing to the player. False positives are also a thing we have to look out for a lot especially in games like Uncharted where it's so easy to create a traversible path somewhere we don't actually want the players to go. 

Sounds trivial  but it's also incredibly valuable to have artists constantly play their levels. It's something that's easy to loose sight of in the heat of production.

And if all else fails it's time to break out the bird poop and yellow paint ;)

- Twitter user Will T Atkers asked:

About how long does it take for a main character to go from concept to final phase?

Andrew: Depends. I'm a strong proponent of getting something in the game as soon as possible, no matter how rough. But from then till someone rips the gold master from out of your sweaty sleep deprived hands it's continuous iteration and polish all the way. It's one of those art is never finished answers. It's important to note that it's easy to reach diminishing returns when you iterate for a very long time. And it's on you to plan and prioritize accordingly. The Pareto principle is a good example of how it usually breaks down: 80% of the work usually takes 20% of the time and the last 20% take the other 80% of time. The last percent come painfully slow, but they also provide most opportunity to grow. Taking something from 0 to 80% awesome is a relatively common skill, but the difference of 1% between 97% and 98% is almost infinitely more valuable. It is literally the "cutting edge" and sometimes takes weeks or months of work to get to.

Jason: In truth, I’ll say about 1 year.  This is strictly for a front and center character like Delsin.  I’ll try to break it down…

First is creative direction needs. What is the purpose of this character? What kind of attitude will they be exhibiting in the game? What kind of arc will this person have? Do they have a disarming or aggressive sensibility?   All of these conversations happen before we throw down any concept art.

Next we hit up concept and begin to sketch things out. Essentially, this process is fairly iterative with our creative direction team until we’re all in a good spot. One unique thing here I have experienced has been a found footage video concept or ‘personality rip-o-matic’.  

Then a fork happens.

On one road, the character team takes this into pre-production to build an in-game proxy model.  The goal is to check movement, scale, and begin to work out the sass of the character through prelim animations.  How do they run? How do they jump? What does this person do when in idle? What does the shape at a low poly look/feel like?

On the other road, we begin CASTING.  Since we use scans and full performance capture, this is very important process to our hero character model pipeline. The creative direction team creates a character description (2 page max) document of our hero. This usually entails basic details like sex, height, weight, and age. Then the remainder of the document explains this person’s motivations, goals, flaws, potential arcs, and habits.  Then we audition a ton of people in hopes that we find someone that can not only bring out the attitude we’re hoping to get, but also have a likeness that is in-line with our visual goals.  In the case of Delsin, Troy Baker fit the bill nicely.

After we cast an actor, we begin the scanning process.  Full body and facial expressions are scanned and sent back to our team.  Our concept and character team will work with the body scan to make any edits that are needed to fit our hero; shorter, taller, thicker, etc.   We take these measurements and send them to a fashion designer who makes patterns to use in Marvelous Designer. Now these two roads have largely converged and we’re onto the next steps.

From here its fairly straight forward. We create hi-poly models of our character based off concept art, bake down and replace our temp PROXY model.  Then comes quality bar iteration through various scenarios. In the past we have 2 scenarios that drive the biggest feedback.

First, the in-game model must be awesome. Since inFAMOUS is a third person action game, the in-game model matters the most. Seeing them running around in the game world with good lighting performing polished versions of those proxy model animations usually highlights a few issues. Sometimes we can’t see them at night so we add better shape or increase value.  Sometimes it’s simply “Delsin isn’t punk enough”, and we add some shiny flare on his vest.

Second, cutscenes will bring the personalities to life. Often cutscenes drive the high end quality bar for us. Unfortunately, cutscenes take a while to make and usually come fairly late in the game.  This is where we iterate on our character specific tech like eyes, skin shading, clothes and hair.

In short, it’s a lengthy process for Hero characters. I wouldn’t be lying if I said a year…  We spend a lot of time ensuring our main character is tonally sound in the marriage between Art Direction and Creative Direction.  For a second tier character we are looking at 2 or so months. 

Brian: Depending on the character, it varies. Heroes usually take longer. Lara took a full year to develop for Tomb Raider 2013 because we were reinventing her. Generally for principal cast models, we spend 2-4 weeks in concept development, hand sculpted realistic characters take 2+ months (High poly, Low Poly bake, textures/materials/shaders) Blend shapes for facial animation and rigging takes another month. Total time 4 months.  Secondary characters usually are created faster, 2 months from concept to finish. Scanning will change these metrics, but a lot more time is spent on preparation like casting, wardrobe, scanning, processing so a fully realistic character scan can get in game in about a month, but if you scan a bunch of bodies and heads at one time, the process is much faster.

Greg: The time that it takes to go from concept to final model depends on several factors.  By working with concept and design the complexity of the character gets defined and the amount of articulation can impact the timeline for production significantly.  Depending on how important of a role the character has will also dictate the time spent in concept.  For us we typically can take about two weeks in concept for one of our aliens in XCOM, but that can really vary and it will stay in concept until we are happy with the direction.  The modeling for something like an alien can take around 4 weeks on average.  With procedural characters, like the soldiers in XCOM it is much more difficult to quantify the time.  A lot of effort is put into developing the systems and how all the parts interact, as well as planning for armor upgrades and customization options. This is really a huge task and quite a bit different from a character that is more self-contained.  Pipelines are also much less linear than just going from concept to model and there is a ton of overlap now.  We typically have the modeler, animation and rigging team involved during the finalization of the concept.  All of these disciplines are able to give input before we go into real production of a final character. Once we are happy at the concept level the model is blocked in and goes to rigging and animation.  Feedback is compiled and applied to the model at this stage. The major point for us is that the model isn’t finished until it is in the game and moving.

- Twitter user Travis asked:

What can you do to keep a cohesive art direction between world assets and FX?

Denis: All my projects so far have been stylized and We went through heavy iteration to find the right style. One initials setups we did on my last project was a small art ready diorama. We made several versions of VFX to see how they'd fit. We also concepted some of the VFX which was helpful. But because of the abstract nature of effects I still consider them the hardest to iterate on. In the end it boiled down to try and error for us.

Combat FVX are a Beast of its own. They have  a lot of requirements like  damagetype, area of effect, faction/monster affiliation and on top it has to resonate really well with combat design, sound and animation.

Having so many pieces that have to collaborate on VFX you have to make sure that everyone knows what the goal is and where we are aiming artistically. Since constantly supervising this process is something you cant afford in production.

Greg: I don’t see these as separate things.  The overall rules for the look of the game directly apply to the effects just like any other element in the game.  Just like a prop in the environment, effects need to be developed in context and not in isolation.  Any effect is an extension of the thing it is attached to, whether it enhances the environment or a character’s ability.

Brian: I always think of them together, FX and Lighting will make every environment come to life.  It's important to ensure FX only dominate a location when they are the star, like being in a level that is on fire, or flooded, etc. But always ensure the style of FX matches the Art Direction of the Environment.  Zelda is my favorite example of stylized FX that match and enhance the Art Direction.

Andrew: Agreed with the gents that this issue is not specific to FX. Maintaining vision and boundaries is crucial with any aspect of art direction. And encouraging people and departments to communicate and collaborate is the timeless challenge of any production. Getting people in the same room to talk and removing middlemen is always the best remedy in my experience.

Technically there are issues with having particles look integrated and leveraging environment data is always a big help. Making sure particles light consistently and accurately under environment lighting and don't need any shader hackery is a big one. Using environment data like surface color or material for the types of squibs or fx to play also add a lot to tying it alltogether. There are cool geo-based particle approaches coming up so having them integrate with the world should become even easier.