Sunday, April 13, 2008

The joy of 2D avatars

I've been looking at 2D avatars lately. It's been a fascinating trip into a wierd little area of game art that I haven't dabbled in before. Like quite a bit of game art, there is a very obvious craft involved in the creation of 2D avatars. It reminds me a lot of the techniques that went into old school pixel art or tile creation. You build your pieces just so according to a very particular set of rules. Align the hand, align the head and voila, the end result look like a unique character.

I ran across a couple classes of 2D avatars that are worth describing. This list is by no means exhaustive and what you ultimately end up using completely depends on the type of game you are making. The different styles can be classified by:
  • Perspective: What view is the avatar seen from?
  • Construction technique: How is the avatar assembled?
  • Animation: How is the avatar animated?

Front view: A frontal view of the character. The benefit here is that the character is often symmetrical which reduces the skill needed to draw a character. The downside is that the character is almost always looking straight out at the viewer. May PlanetCute characters are a good example of this perspective.

Partial side view: A partial side view of the character where the character is rotated 45 degrees to the left or right. The example above is such a character. This character can interact with objects in the environment, but with subtle (and cheap!) animation of the eyes, look at the viewer. You can make avatars with this perpective move left, right in a believable fashion by simply flipping the avatar. Climbing ladders is less than compelling. :-)

4 (or 8) directions: For each of the cardinal directions you draw a new version of the avatar. The characters in Diablo are a nice example of this style. Having multiple directions is more realistic and allows you to show the general direction that a character is pointing. However, it is again more expensive. If you have an interchangable item on the character, you need to draw 3 different versions for 4 directions and 5 different versions for 8 directions. This multiplies your art expenses.

Isometric: This is similar the partial side view, but usually seen a bit more from the top. This is almost always done with multiple directions.


Interchangable parts: Most avatars are made out of interchangeable parts. You can swap out a shirt for another shirt. This allows for a vast range of different characters, but They all tend to be built from the same basic mold. Luckily, so does most of humanity, so this system tends to work well for humans.

One piece: Whirled and most single player games uses characters that do not have interchangable parts. All the animation is built into the character. The upside is that you can have great unique animations that really fit the character. Your dragon can breath fire and your balloon beast can float merrily along. The downside is that you need to make unique animations. This is expensive.


No animation: This is the simplest and is quite common with web-based games. If you can get away with it do so!

Cell animation: The whole avatar or the pieces of the avatar are animated by flipping through a series of frames. This tends to be a specialized skill and good 2D animators are hard to come by. Cell animation is highly evocative and has been the animation technique of choice for ages. The downside is that if you are using interchangeable parts, each part needs to be animated through all the possible animations. This means that the number of animations that your avatar supports is likely to be small since you don't want to bloat the cost of creating each item.

Vector animation: Each piece of the avatar is mapped onto a simple 2D vector rectangle that can be smoothly rotated, scaled and squashed. Add a simple skeletal animation system (the foot bone is connected to the leg bone which is connected to the hip bone) and you can do some reasonable effective animation. The characters in Book Worm Adventures are a great example of this style. If done correctly, this method lets you animate just the base skeleton and swap in parts as desired. The upside is that you can have a huge range of animations with the same basic art assets.

Little lessons learned

Optimize for ease of object creation, not richness of animation or immersiveness: If you are going for virtual item sales, your incremental profits can be broken down to # of items sold * price of items - production cost. You want lots of item variety and you want to keep your item cost down.

With this in mind, the format of your avatar begins to matter a lot. Animation is expensive and makes object creation even more expensive. Multiple directions for your avatar are cool, but are they really worth increasing your content costs by 500%?

Most items are made out of at least two pieces, not one
: When you build hair for a character, you have a section of hair that goes behind the avatar's head and a section that goes in front. I've been sketching it all out as a single piece and then chopping it up as needed during the cleanup stage.

Use flat shading or indistinct light sources if you are using vector animation: Since your pieces can be rotated in all sorts of directions, highlights will often look strange when rotated.

The number of slots in your avatar represent sales opportunities:
A character composed of a head and torso presents very little opportunity for players to customize their look and feel. After purchasing a couple of items, they are done. By allowing for tiaras, jewelry, wings, thought bubbles and other items, you win by creating additional sales opportunities. The player wins by having more ways of making their character unique.

Style matters
: I dress like the guy in The Fly. My closets is filled with row upon row of identical pragmatic clothes. I wouldn't know the difference between a cardigan and a camisole if my life depended on it (I actually had to look it up.)

Yet many avatars, especially those in online games, are ultimately about fashion and style. The cut of the fabric is important. The patterns matter. The colors...don't even get me started on the colors. It is no surprise that some online game companies (like StarDolls) build up such an expertise in fashion that they are launching their own real world clothing lines. So I've been reading women's fashion mags. It's a whole different world out there.

Next Steps
I'll continue dabbling with these sketches. My character drawing skills are not the best, so at the very least this will be good practice.

Which style should I use? I'm leaning towards a side-view avatar with interchangeable parts that uses simple vector animation. The cost of production is low and the style seems to be the best fit for some of the game design ideas I've been mulling over. The current template has spots for custom headgear, a head, eyes, mouth, nose, top, bottom, feet and items held in left or right hand.

Here is one last sketch.
Comments and critiques are welcome!

take care


Wednesday, April 9, 2008

A Services Strategy for Casual Games

Gamasutra posted up an article that has been bouncing around in my documents folder for a little while. The original title was "A Services Strategy for Casual Games", but the new one is a bit more punchy.

One response that I've heard quite a bit is that portals will never allow user data to be released back to developers. This is quite true for most established portals that have traditionally focused on selling packaged goods online. However, middlemen adapt and markets flow around stupidity. More sophisticated variations on sites like are bound to emerge. If a dozen portals don't want your business, find the one that does. Given time and a exclusive supply of successful games, they'll grow into a bigger fish that can help feed your team.

The portals are engaging in a kneejerk reaction to changing business models. In the long run, do they really think they can keep customer data away from developers when the games that players want are online services? Such companies just end up being a roadbump in the way of progress. A portal that gets irritable about giving up customer data guarantees that their cut of the pie is zero. This is their loss, not yours.

take care