Eric Kimberly

He’s leaving us to go join Sledgehammer and we wish him the best of luck.

Below is just a small sample of some our favorite pieces created by Eric while he was here.

Marius Millar

His artwork is amazing and he is really helping out.

The creation usually begins with a piece of concept art, and figuring out the points of interests of the character, his or her personal narrative, and all the little trinkets, clothing and pieces of history that will populate their body.

All of it will eventually be created in 3d or captured within the texture information. Modelling and texturing are two very different sides of 3d, but they go hand in hand in game art. One is usually done after the other, due to the fact that the finished game model of a character, or anything 3D, requires UVs; which we will dive into later.

We have to start with the modelling!

As with most games with normal maps (which most games have now a days), characters are built from the ground up. The low poly is the final product, but it uses a normal map to create the illusion of higher fidelity. So we get our normal map from a high polygon sculpt.

And when I say high, I mean high. There are a lot of sculpting programs out there, like Sculptris or Autodesk’s Mudbox, but I prefer using Pixologic’s ZBrush because of the pure power it has in rendering a large amount of polygons. Our game ready character models are made up of about 10,000 four sided polygons (or 20,000 triangles), and ZBrush allows you to have millions of polygons in your scene! Millions! Maybe even billions if you have enough RAM. But I definitely didn’t need that much for these high resolution sculpts.

I usually start my models with a generic female or male base mesh, pull limbs and other body parts into proportion, and then start building up from there. I don’t really do anything fancy; its just basic building up forms and sculpting. Once I’m satisfied with the body sculpt, I’ll start grabbing basic clothing meshes I made and add them into my ZTool, and sculpt those into place. ZBush’s subtool feature is super powerful, and thanks to things like insert meshes, and the super handy ZRemesher feature, sometimes you don’t even have to leave ZBrush to make the things you want.

Here’s a timeline of how I went through the development of Chiyo inside of ZBrush.

This is pretty much how most game res characters are built these days. When you create for yourself a high resolution sculpt, there are so many opportunities for you to grab all sorts of information when you take it to the low poly process. But that’s not super important right now, what’s important is getting the shapes and forms of the character developed.

Making the Game Resolution

When we finally do have our finished high resolution sculpt, and everyone likes it, I begin the process of creating the game resolution mesh that will be used within the game engine itself. Since we have given ourselves a pretty decently sized budget for our characters, I tried to not to sacrifice the detail or silhouette of our characters throughout the process.

It’s a process that  mainly involves drawing a shell on top of your sculpt. It can definitely be a time consuming process but its important in creating proper animatable topology that our animator can use so the mesh deforms correctly when its animated.  I create my low poly model using Maya’s new modeling toolkit feature they introduced in 2014.

This also makes the mesh usable within the game engine. Like I said before, your meshes inside of zBrush can be upwards of millions of polygons, and that is completely unusable inside of Unity, at least for just one game object.

You do this for every part of your character, and you’ll end with a whole bunch of low poly pieces, that are now ready to be animated and textured!

The UV’s

This is probably the most unglamorous part of our character breakdown because its pretty technical, and its hard for a lot of people to wrap their head around.

Imagine a bear skin rug. Okay now imagine that with a game resolution face.

Basically we are laying out our character pieces and shells so that they can be painted on. It’s an oddly satisfying feeling when you pack them all into a single uv sheet.

Now that the boring stuff is out of the way, its on to the fun stuff!

Texturing

Wait just kidding. All the boring stuff isn’t done yet. We still need to make our normal map! This is still technically part of the texturing process, because we are going to use a system of projecting the high resolution details onto the low poly mesh we made earlier. All of this is nearly automated inside of your favorite tool (I use xNormal), and you can get nearly any kind of texture map you want out of this process. Your normals, your object space normals, your ambient occlusion, your curvature, your cavity, your height map, EVERYTHING!

Basically a normal fakes details that would be too small and would require a ton of extra geometry to show up on your character. Next time you are playing your favorite game, go take a look at a wall! The illusion of the game world around you will be shattered once you find out its not actually made of bricks, its just a picture of bricks that are being faked with some depth. The same goes for characters! Scars, wrinkles, and clothing details are usually all done within a normal map.

The Actual Texturing Part

Now we can actually begin texturing! Usually this process is done within your favorite painting program, usually Adobe Photoshop. A lot of texturing relies on a lot of masking details, generating details, etc., and Photoshop is perfect for that.

Layers are my best friend. We start with our basic ambient occlusion map, give it some basic color and build it up from there. It’s a very fun and rewarding process, because you can see the effects of what you did almost immediately on the model.

Hand painting in details goes a long way to to bring out features of the face and to make details of the clothing pop. But wait, this is just color information! What about actual material definition? How do we actually make the steel look like steel and the gold look like gold? That’s when we turn to our specular and gloss maps.

Materials

For Head of the Order, we are utilizing Shaderforge, and it gives us a ton of options for creating our character shaders, and most importantly, allows us to do a lot with our specular and gloss textures. Defining our materials is super important so that the character doesn’t look like plastic when we bring her into the engine and just apply a flat specular to everything.

Using the default unity bumped specular material doesn’t give a ton of control over what value you can give your spec maps, how glossy things are, and how much color is in your spec maps.

The color information we made previously is also known as a diffuse map. What we want now is our own authored specularity and gloss maps. Specularity helps the shaders know what colors are reflected off the character when he or she is lit, and gloss helps to determine how sharp that lighting it is, or how rough and diffuse it is. We’ll use Chiyo’s clothing for this example

If just a number is applied, it doesn’t look right, which is why we need to author the maps ourselves, so that it will actually look and feel like metal. It’s a very rewarding process.

With that, the character is pretty much complete! The specular and gloss maps sure do make a difference. Our final character comes out to 22,545 triangles, and uses two sets of 2048 textures. Look for this character and the many others I’ll be making for Head of the Order when it launches later this year.

After this, I sketch. I try to find a pose that makes it easy to see every detail. This is important for when I hand the image off to the 3d team. My first pose looked too much like a zombie, so I tried to give him a more natural stance.

After discussing the sketches with the team, I choose which aspects to keep and which to discard, and then generate a clearer line drawing. Linework is important to me, and I enjoy the gradual progression of clarity it affords. It allows me to get a feel for the costume and accessories of the character and decide if their is anything that just isn’t going to work. I also use this time to mentally plan for the painting and think about color choices.

Next I block in colors by creating a layer underneath the linework layer.

After this, I turn the opacity way down on the previous linework layer, and create another layer above that. I use the rough pencil tool to go over the lines one final time. I like to keep the lines as clean as possible, and have them mostly be on the outer edges of the drawing.

Now it’s time to paint! I again create a layer under the linework to paint. I typically keep each item of the character on its own layer – like the scarf, the character’s skin, the shirt, etc.

I usually paint the base color of the item first, the shadows second, and the highlights last. I often take advantage of the color burn/color dodge tools for this, but I keep them on a layer above the paint layer until I’m sure it’s how I want it to be. I don’t use many layer styles or effects, I prefer just painting. Color and lighting are some of the most challenging things for me, and I’m always working toward improving my skills in those areas.

I’ve run into a lot of artists who seem to feel that working off reference is considered “cheating”, as though they should be able to pull an amazing drawing or animation right out of their heads with no trouble at all, but the opposite is generally closer to the truth. Working with a great deal of reference as often as possible also creates a mental library of poses one can call on whenever they need.

The same goes for animation.

I generally start an animation by shooting reference or looking some up online if the movement is particularly complex or non-humanoid. Sometimes it’s fun to just futz around in Maya and animate on a lark without reference, but if you have a specific goal in mind, it will save you a lot of time. For game animation, I’m always sure to get several different angles of the same movement and a few different takes.

Once I’ve got good enough reference, I’ll double check my frame rate in Maya and start setting up the key poses based off the video. The key is to find the poses that comprise the essence of the movement you’re looking for, and be ready to push and pull keys around to correct the timing. I was originally taught to use stepped curves when laying out my keys in order to check the timing, but in practice I generally prefer splined curves—stepped curves can be jarring to me and I find I have an easier job with timing if I go with splined. Throughout this process I’ll be checking on the graph editor to adjust the tangents. After that, it’s all in the noodling and finagling to get the movement just right.

Another important part of the process is recognizing that Maya can take care of a good chunk of the in-betweens for you. Animating in 3D doesn’t work the same way as 2D; you don’t need to tell Maya what to do in between every frame beyond correcting the curves if they need it. Every time I create a new animation, I always rediscover how less is more when it comes to keyframes. Half the time if I have a movement issue, I’ll end up deleting a frame or two and that will solve it.

Another important note about art in general: sometimes it’s best, and easiest, to scrap a piece and start over than to try to salvage it. Don’t be afraid to start over if whatever you’ve made isn’t making you happy. You’ve already made the good parts of it once before; undoubtedly you can recreate it.

Here is an interview of our Art director Melissa while we were showing at GDC 2014.