I recently had a client who had been reviewing different avatars tell me that they all seem the same. Granted, this person is in purchasing, not in learning, but the comment does bring up an interesting issue. What makes a good elearning avatar?

We've learned by experience that characters that are too photorealistic are creepy. That's because the rigging and animation always fall way short of the visual features, causing a distracting Max Headroom kind of dissonance.

Simpler characters compliment learning, rather than distract from it...and yet, too simple also gets back to being a distraction. So not photorealistic, and not the rather crude talking heads you sometimes see. And while there is certainly a place for 2-D characters, we prefer 3-D, for most applications.

Just as the voice can convey important emotional cues that add significant depth to a message, so can body language. That's why we like to see a character that can walk and gesture, and of course lip-synchs with the voice. 

When it comes to production, a good avatar is easy to work with and keeps ROI in mind. That's one of the main advantages to the Codebaby characters we use. We can change clothes and hair pretty easily...and they come already rigged, with a broad selection of movements and gestures for us to plug in.

Bottom line? The avatar is there to help create a more human experience. They make practicing a learning objective much more real and powerful. You need them to do that job without being a distraction, and without causing you more work than the project is budgeted for.