Anthropomorphic Avatars

My first anthro attempt from 2009-ish lololol:

My anthro prototype from 2016 (I’m too lazy to add hair styles and finish the male):

Fresh memes:

Characters are limited by your animation and rigging tools. I can move joints and see the animation change in real time. Then I just copy&paste the folder containing the animation into my game.

I can feed arbitrary functions that return cframes to my animation system. This enables me to implement inverse kinematics,

Without detailed UV maps or mesh skinning, anthropomorphic characters are extremely difficult to implement in a non-cringy way. Face animations do help a bit:

If this was instead a bearded dude wearing armor that had HD spec/bump maps, I think developers would have had a very different reaction to avatar 2.0:

Also there will be problems in animations when transforming the character’s skeleton variably between normal r15 and anthro. Some animations might depend on the position of limbs in space, and distorting the skeleton can cause the end-points to be off. Roblox should be very careful moving forward with this; Good luck supporting this character slider in 5 years when you want to implement fully articulated hands to keep up with what developers are creating. Don’t get me wrong, the demo is great, but every change to the avatar results in more complexity and edge-cases for developers to deal with long-term.

The animation system and server-client model for characters/animations/physics needs to be perfect. There are cases where animation packs influence a player’s ability to complete an obstacle course, in which case the animation should be purely graphical/cosmetic; There are also cases where a developer might want to make a crouching animation and have them fit under obstacles, in which case the animation needs to influence physics.

At what point should a developer use constraints instead of animations? Realistic physics is a huge part of Roblox’s vision, and perhaps we can learn something from animation systems like Euphoria. (from 2008):

Animation packs and graphical cosmetics are very profitable, but animation packs like levitation don’t truly fit roblox from a physics standpoint. My games make a clear distiction between graphics and physics for this reason:

• Complex graphics can be generated client-side, instead of by the server
• Physics is stable and overhead on the server is low
• Animations can be creative (backflips, dancing, flying)

In 10 years we might create flying animation packs by applying forces to a character’s physics assembly to keep them in the air, or we can just divide graphics and physics, idk.

I think the best course of action from pre-r15 would have been to do automatic mesh skinning on shoulders/hips to make it look like there are elbows/knees (basically treating the tip of the hand or the bottom of the foot as an inverse kinematics target.) There might not be an elbow joint in the animation skeleton, but it can be implied using IK. This would have preserved the platform’s simplicity while spicing up the graphics without any backwards or forwards compatibility implications. Although when it comes to actually developing a game, custom complex skeletons are necessary, and r15 makes anthro characters more accessible to new devs.

Edit: Fixed dead gyazo links.